1. Introduction

Archaeoastronomy, the combined study of ancient cosmology and archaeology, has been characterised as divided between its European 'green' wing that narrowly uses statistical techniques, compared to the 'brown' archaeoastronomy of America where a 'rich ethnographic and ethno-historical record (...) relegate[s] statistical arguments to a secondary supporting role' (Ruggles and Saunders 1993, p.15). This suggested division within the discipline is overdrawn and cannot explain present problems in the discipline. While surviving cultures and rescue anthropology have been able to celebrate and record some of the precious messages of Amerindian cultures, it remains the case that the archaeoastronomy of the two thousand year old Newark Earthworks, Ohio, for example, have no known cultural descendents. Scholars in this area of 'brown' archaeoastronomy have had to resort to exactly those statistical techniques usually associated with the 'green' Europeans (Hively and Horn 2006). Contrarily, recent statistical findings in Britain for an abundance of lunar alignments in prehistoric monuments (Ruggles 1999) converge with the anthropological data of many cultures around the world which suggest that they have entrained their ritual cycles with those of the moon (Knight 1991). Yet although we can reference modern low latitude hunter-gatherers such as the !Kung and Hadzabe (Power & Watts 1997) and high latitude reindeer herders such as the Saami (Ahlbäck 1987), all of whom synchronise their economic and ritual life by lunar and lunar-solar phases, or Palaeolithic lunar calendar sticks (Marshak 1972), simply accumulating confirming evidence will not allow us to leap over the inductivist dilemma that these ethnographic exemplars are not the cattle herders who also hunted and occasionally planted who actually built the monuments of the late Neolithic British Isles (Thomas 1999). Some other method beyond counter-posing the ethnography of recent cultures and the statistics of prehistoric monuments is required to address the problem shared by both American and British archaeoastromers - that we have no direct ethnographic data of the builders of prehistoric monuments.

The distinction between 'green' and 'brown' archaeoastronomy is also not useful when we try to understand the achievements of the statistical method to the recent history of the discipline in Europe. Over the last 40 years European archaeoastronomy has successfully transcended the shortcomings of the work of Hawkins and Thom (Heggie 1981; Ruggles 1999). It has rebutted claims for the over-interpretation of monuments' alignments by establishing rigorous field work methods and the statistical analysis of regional groups of monuments, so establishing that many prehistoric monuments have 'astronomical' alignments intentionally built into their design. In Europe these gains have been made in spite of a largely sceptical archaeology establishment. However, such has been the strength of this scepticism that European archaeoastronomy has settled into a narrow routine seemingly in an effort to gain wider acceptance. While providing a way to reject the null hypothesis, the statistical approach cannot be used for testing intentional alignments in unique monuments such as Stonehenge or the Newark Earthworks. This leaves the field open for other disciplines to monopolise such unique and defining monuments at the expense of any archaeoastronomical input. Yet at least four other methods are available that can interrogate individual monuments. Monte Carlo modelling constructs a virtual population of randomly generated alternative 'unique' structures which allow statistically testing for the null hypothesis. This procedure generates a set of pseudo-data assuming the null hypothesis is true, and by comparing the actual data against the simulated data we can test whether chance or intentional design can explain a unique monument's alignments (Hively & Horn 2006; Ruggles 1999). A quantified landscape phenomenology looks at the landscape context of any monument as a region of alternatives, so facilitating tests for whether the actually chosen site exhibits any particular portfolio of properties which may include an 'astronomical' dimension. To do this we systematically search for all the logically possible ways a monument complex could have been located in its local landscape.

By comparing the phenomenological and qualitative experience of each simulated alternative route and design against the actually chosen arrangement, we can quantitatively test whether there emerges a suite of embodied experience not available in any other location (Sims 2009). Isolating detailed features of a monument that are unexplained by other hypotheses allows testing for an astronomical property (Ruggles 1999; North 1996; Sims 2006). And virtual modelling of monuments within accurate computer models of landscape and skyscape provides another test (Macdonald 2007). Together with the statistical analysis of regional groups of monuments, these five methods constitute a significant battery of techniques to test the null hypothesis for intentional alignments for both regional groups and individual monuments.

The discipline has also become characterised by an inclination to seek single axial alignments for any one monument, and to define the builder's cosmology as limited and defined by that single alignment. Thus a regional group with axial alignments on winter solstice would be seen as having a distinct cosmology to another regional group with summer solstice alignments (Hoskin 2001). This is in spite of many well attested cases of complex monuments exhibiting a grammar of combined alignments arranged in parallel, transverse and reverse pairings (Ruggles 1999; North 1996; Hively and Horn 2006). Therefore unique monuments like Stonhenge and Avebury stone circles that possess alignments on both winter and summer solstice settings confound this limitation (North 1996). And even though one of the most impressive field work reports and statistical analyses ever conducted in archaeoastronomy has found a preponderance of lunar alignments (southern major and minor standstills in particular) in five regional groups of prehistoric monuments in the British Isles (Ruggles 1999), both the author and the discipline are reticent to submit lunar data sets to intense investigation. This is in keeping with a deep assumption within archaeology that such is the complexity of the moon's horizon properties compared to those of the sun, that farming cultures just emerging out of foraging lack the sophistication to design monuments with lunar alignments. This view is contradicted by that of anthropology, which sees hunter-gatherers as fully human, as 'sophisticated' as agriculturalists, and who use lunar cycles to time their ritual life (Knight 1991, Sims 2006). "It is evident from this discussion that a serious problem in studying cultural astronomy is the lack of a rigorous methodology for combining evidence from ...[these] three main disciplines..." (Ruggles & Saunders 1993, p. 15). However, in the interests of carving out a professional discipline acceptable to academe, present archaeoastronomical practice tends to focus on statistical methods to test groups of monuments reduced to single axial alignments on the sun.

Science should not be limited to the socio-political pressures of institutional acceptance. We can raise our sights and include in our aims simultaneously testing models other than null from a number of disciplines that predict different 'astronomies' for ancestral cultures. For example a recent palaeo-anthropological model of cultural origins suggests an initial situation in which coalitionary strategies of mega-fauna hunters would have maximised their fitness by entraining their social and ritual life with the phases of the moon. The subsequent collapse of these coalitions with the mass extinction of big game at the end of the last ice age would have impacted upon their ritual systems, and the consequent degradation of the original lunar template should be traceable through a series of transformational adjustments. In particular this model would predict that while power was originally mobilised at dark moon by female coalitions able to synchronise their cycles in seclusion rituals, later male monopolisation of ritual power would have to accommodate and displace dark moon symbolism onto solar or other cycles (Knight 1991, Knight et al. 1995, Sims 2006, Sims 2009). By the late Neolithic the prediction would be that monument builders would have complex cosmology in the service of an elite to both appropriate and subvert a lunar cosmology once held in common. This model and its predictions is precisely different from recent archaeological models, which sees the first 'agriculturalists' as emerging from forager 'primitiveness', and in which the earliest monuments would at most have simple alignments on the sun. By selecting these archaeological and anthropological models to interpret monuments, and testing them alongside the evidence of archaeoastronomy, we can suggest how robust datasets drawn from these three (or more) disciplines have a limited number of combinations, which in turn allows only a very few reconstructions of an 'emergent' cosmology (Sims 2009). This paper demonstrates this argument through examining claims made by some archaeologists for the approach made by the West Kennet Avenue into the Avebury Henge, in Wiltshire, England.

2. Archaeology and the West Kennet Avenue approach to the Avebury Henge and Circle

The West Kennet Avenue is one part of the unique Avebury monument complex built during the third millennium BCE that connects the Sanctuary wood and stone circle with the Avebury henge and circle (Fig. 1). The Avenue was composed of about 100 pairs of parallel stone pillars, of which those remaining after centuries of abuse have an average height of 2.26metres arranged in quadrangular settings 14.7 by 23.2 metres apart (Sims, Field Notes).The excavations and restorations carried out by Keiller and Piggot showed that this part of the Avenue was built in series of ten straight sections, not in a smooth serpentine shape as suggested by the eighteenth century antiquarian Stukeley in Fig. 1 (Keiller and Piggot 1936). Where stones were missing, they placed concrete markers above the excavated stones holes where they had once stood, so providing a record of the northern section of the Avenue. Stones and markers are identified by numbered pairs 1-37 going south on leaving the Avebury henge, and by row 'a' on the east and 'b' on the west. Paradoxically Keiller's plan survey of this section of the West Kennet Avenue shows it heading away from the southern entrance of the henge from pair 13 to 6, while from pairs 6 to 1 it seems to repair the 'error' by an awkward zig-zag route to then connect with the southern entrance (Fig. 2). Recent archaeological commentary on the Avenue has suggested two interpretations for this convoluted approach route. Burl claimed that this was a mistake of the prehistoric builders in starting the Avenue at both ends but failing to anticipate an accurate direction for each section to join up (Burl 2002). Gillings & Pollard argue that Keiller's excavation plan is a mistake and re-excavation will establish a more direct route for this section of the Avenue (Gillings and Pollard 2004, p. 78).

Burl's suggestion of the builder's poor route-making abilities might be taken seriously if the join in two sections took place in the middle of the 2.4 kilometre Avenue, but it is unconvincing when the 'poor join' at stone 4b is just 30 metres from the southern entrance. If it were a mistake, then it cannot explain why elsewhere in the Avebury monument complex are displayed highly accurate pre-planned features (Sims 2009). Lastly, an earlier antiquarian of the seventeenth century, John Aubrey, recorded how the other end of the Avenue connected to the western entrance of the Sanctuary in exactly the same dog-leg design, the same flat sides of the stones in line with the Avenue route, and the same device of using a change of slope in the landscape in the final approach to the Sanctuary as in the northern approach to the Avebury henge and circle (Fig. 3). However, Burl's view of the builder's logistical incompetence is consistent with the archaeological assumption of farming revolution theory that they were 'howling barbarians' (Atkinson 2003) just emerging from the primitivism of foraging.

Pollard's suggestion that Keiller's excavation record is inaccurate is also suspect. It is true Keiller made some mistakes in his record of the West Kennet Avenue. He placed a concrete marker at position 30b, where no stone had ever been placed, and he failed to place a marker at position 4a where a stone had once stood (Smith 1965; Sims forthcoming), although these are errors Pollard would find hard to accommodate in his theory of avenues. Pollard sees Avenues as lithicised commemorations of pioneer ancestral pathways into a region, and considers it improbable that those Mesolithic foragers would have taken such an indirect route as Keiller identified. It also suits the sensibilities of the modern tourist board for the monument (the National Trust) who have mowed a short cut along this section of the Avenue which ignores stone pair 6 in a streamlining of its actual route (Fig. 4). But this is a theoretical, not empirical, case for challenging the zig-zag Avenue route near its connection with the southern henge entrance which would fall if we can find another theory which can explain the archaeological evidence of Keiller's site excavation.

3. Anthropology and the Shift from Brideservice to Brideprice

We have two models from archaeology, the primitivist and memory models, which deny the evidence of one of the most accomplished practitioners of the defining method of archaeology – site excavation. Anthropology would be concerned that in the transition from Mesolithic foragers to Neolithic 'farmers', key politico-economic changes were being ignored by this archaeological view, and these may account for this paradoxical property of the Avenue. The switch from hunting and gathering to cattle pastoralism involves hunting brideservice being subverted by cattle brideprice. In pastoralism a man gains a permanent wife with a payment of cattle which would substitute for a lifetime's hunting services to her kin (Aberle 1961; Douglas 1969; Holden and Mace 2003; Jamieson 2010; Murdock 1949; Richards 1950; Schneider 1961). Women are now 'wedlocked' and men are divided by differential cattle ownership. Or as Aberle put it: 'the cow is the enemy of matriliny' (Aberle 1961, p. 680). Thus, whereas farming revolution theory sees a rise in 'civilisation' from foraging to agriculture, anthropology sees a socio-political reversal in marital and economic relations combined with an advance in technology. Archaeoastronomy provides a method to test hypotheses generated by these different models and a procedure to integrate elements of each model that survive critique. Archaeological models emphasise single axial alignments on the sun or none at all. Anthropological models predict lunar symbolism for hunting cultures that would be gradually undermined by solar symbolism with the beginnings of pastoralism (Sims 2006). It should be possible to observe traces of either paradigm in the paradox of an Avenue that until the last moment heads away from its destination.

4. Unexplained Design Features of the West Kennet Avenue

We can begin by respecting the details of this section of the Avenue by looking for aspects of its design which are unexplained by the current archaeological models. From pair 13 to 7 the Avenue is 10 straight and takes us downhill towards the henge in a line that is a tangent to the outer south western bank of the henge which continues on to the summit of Windmill Hill (Fig. 1). Just at the point where the modern tourist footpath veers away from the Avenue in a modern shortcut, stone pair 6b's position continues veering away from the southern entrance by occupying the lowest point in this section, only to require a sharp turn to the east uphill to pair 4, followed by another sharp turn this time to the north to pair 1 and so into the henge (Fig. 2). A route that loses height to then require immediately regaining it is not what we would expect of Mesolithic foragers, just as tourists today seem to agree by taking the modern shortcut! However, such a strategy is perfect for lowering the eye of the observer processing along the Avenue. Passing stone 4b, which is the only remaining stone from markers 1-12 along this section of the Avenue, this flat quadrangular slab of a stone (Fig 4) is aligned in line with a route consistent with the zig-zag alignments confirmed by Keiller's excavation (Fig. 2). In all these respects - the Avenue as tangent to the outer henge bank rather than entrance, a route that crosses contours to change the altitude of the observer's eye, and the flat sides of a surviving stone confirming these layouts – Keiller's record of the northern terminus replicates all of these properties that we find in Aubrey's record of the southern terminus of the Avenue. Unless Pollard has evidence that all the concrete markers from pair 12 to 5 are incorrectly placed then the conclusion must be that Keiller's record is correct and needs to be interpreted with some model that transcends the limits of the primitivist and memory models. An alternative model is suggested by the properties we have just noted. A circuitous route which manipulates the eye height of the observer is simultaneously altering the altitude of the surrounding horizons. Having set aside two archaeological models we can now test with archaeoastronomy whether horizon events coincide with the arrangement of stones along this restored section of the Avenue.

5. Method

We have no knowledge of how the builders of the Avenue might have aligned paired and adjacent stones along the Avenue. For those archaeoastronomers wedded to a 'Thomist' expectation for highly accurate alignments (Thom & Thom 1976; see Heggie 1981 and Ruggles 1999 for critique) this poses a problem. The average mid-width of the surviving stones is 1.7 metres and pairs are placed on average 14.7 metres apart and diagonals 27.5 metres apart. This allows maximum average ranges across alternative sighting alignments of 11° for paired stones and 5.5° for diagonal stones – huge ranges for those accustomed to plotting alignments accurate to fractions of a degree. However, if you accept an 'ethnographic' (Ruggles 1999) or 'religionist' (North 1996) motivation for ancient sky lore, then this large range of sightings over the tops of stones when viewed from adjacent stones is an advantage for constructing the artifice of observing cosmic bodies entering or leaving the stones at settings and risings. Seen this way, the stones can be constructed as 'portals' for the passage of solstice suns and standstill moons between the heavens and the underworld. In keeping with this ethnographic logic a range of 5° to discriminate any horizon event of the sun or moon along the top of any stone when viewed from an adjacent stone is the minimum accuracy required to capture this effect (Sims 2010). While farming revolution theory would find it difficult to accept the high fidelity alignments predicted by the Thom model of archaeoastronomy, there is no reason why it cannot accept this 'religionist' understanding of prehistoric sky lore. We can classify all of the ten logically possible adjacent alignments for any pair of stones as shown in Fig 5. Azimuth field sightings were made in both directions with a compass accurate to half a degree, and horizon altitudes measured with a clinometer. Field work was repeated three times with different observers to check for recording errors. In addition a virtual model of the Avenue using Keiller's site survey plan was built independently of this fieldwork by MacDonald, integrated into an accurate virtual landscape using Ordnance Survey topographic data, all combined with an accurate and realistic moving skyscape that could be set for any date between 4k BCE to the present (MacDonald 2007). An example of one of the many alignments captured from this computer simulation is shown in Fig 6, and the field data results are shown in Table 1 (and see Appendix 1).

6. Findings

While this paper concentrates on the Avenue properties between pairs 7 and 1, the 167 lunar-solar-cardinal alignments shown in Table 1 between pairs 37-1 are far more than can be accounted for by chance alone (Sims 2010; Appendix 1). It is also clear that there is a good 'archaeoastronomical' reason why the approach route of the Avenue does not head straight for the southern entrance, but veers to the north-west at a tangent to the outer henge bank – it's line is dedicated to align on the summer solstice sunsets between pairs 12 and 7 with a switch to the northern minor standstill moonsets between 7 and 6 (orientations 9 and 10). The subsequent zig-zags between pairs 6 and 1, rather than being 'awkward' or a mistake are also made to maximise further cardinal and lunar-solar alignments. Interestingly, at the junction of pairs 7 and 6, the main point at which the Avenue changes direction towards the southern entrance, there are combined cross cardinal alignments to the west and the north. These cross-cardinal alignments match the builders' gender inflected burial practices in which gender is demarcated by cardinal alignment, with an 80% emphasis on male burials all orientated to the north (Tuckwell 1975). It is also interesting that while the main direction towards the southern entrance by way of the Avenue concentrates on summer solstice sunsets up until pair 7, from then on all remaining adjacent pairs emphasise lunar alignments on northern minor and major lunar standstill moonsets into the henge. This lunar-solar combination of summer solstice sunsets and northern standstill moonsets, predictably and invariably generates a synchronisation of dark moon at summer solstice – exactly what would be predicted by a lunar governed ritual system based upon dark moon seclusion rituals transposed onto a solar timescale (Sims 2007). For those processing north along and outside the parallel rows of Avenue stones, they would have seen a changing vista of stone combinations of first summer solstice setting suns setting into their tops followed by, twice every nineteen years, the setting minor and major standstill northern moonsets. Fig 6 shows a computer simulation of this effect. At that part of the Avenue when its zig-zag route disallowed a lunar-solar alignment across stones, they contrived to manipulate their position to construct an alignment on north – a direction that they also used to align male corpses. Solstice suns and standstill moons setting into stone tops and the northern centre of the heavens would be seen when travelling alongside the Avenue when moving towards the Avebury stone circle, and risings when moving towards the wood and stone monument of the Sanctuary.

7. Conclusion

The 'astronomical' properties we have found for the West Kennet Avenue approach to the Avebury henge and circle are exactly what would be predicted by an anthropological model in which a male-dominated cattle herder society is appropriating and subverting a lunar-governed ritual cycle onto a lunar-solar cosmology. This model can therefore provide an interpretation for the excavation findings of Keiller and Piggot which modern archaeology cannot. Where once foragers had naturalistically entrained their rituals with lunar cycles, adapting to mega-fauna extinction through the technical advance of domesticated cattle as moveable property, gender and economic relations are now characterised by compulsion and inequality. Now ritual specialists had to construct 'the pathways to the gods' to keep a connection with their ancestral beliefs while simultaneously undermining them. This model both explains the findings of archaeoastronomy and at the same time integrates those findings that remain from archaeology and anthropology. In this multidisciplinary integration of data sets, it is archaeoastronomy that is the keystone discipline in the intellectual arch. A unique Avenue aligned on sun, moon and cardinals whose route goes the 'wrong' way for archaeology may point the 'right' way for the future of archaeoastronomy.

Acknowledgments: I would like to thank Sasha Stephens and Jorg Endelman for assistance in conducting the field work for this paper, and to the anonymous reviewers and editors for their comments and advice.