1) Rice. PREHISTORIC FORTIFICATION SYSTEMS OF NORTHERN PERU PRELIMINARY REPORT ON THE FINAL SEASON JANUARY-DECEMBER, 1980 Theresa Lange Topic John R. Topic Department of Anthropology Trent University Canada oo May 26, 1982 -In 1980, the Forts Project completed its survey of fortifications in its survey zone in Northern Peru. During the summers of 1977 through 1973 the project had carried out survey and mapping of prehistoric fortified Sites in the western sierra of the Department of La Libertad. From its headquarters in Otuzco, the project had concentrated its energies on the upper drainages of the Moche and Viru Rivers (see Figure 1), During the calendar year 1980 the project was based in Trujillo and the emphasis of the project was on sites at lover elevations. Substantial programmes of survey and mapping were carried out in the middle reaches of the Moche and Viru Valleys, excavation was done at key sites in the Moche Valley, and reconnais~ sance was made of selected coastal fortifications from Jequetepeque south to the Chao Valley. ‘The project in general was heavily oriented toward survey, with the emphasis on fortified sites, defensively located sites, and isolated defen- sive features (e.g. walls). In the Otuzco area our survey was the first formal archaeological research to be done; thus it proved valuable at times to include unfortified sites in the survey to aid in the construction of the ceramic sequence for the zone and to obtain more complete information on settlement systems and their changes through time. In the better known middle valley and coastal zones, the emphasis was exclusively on fortifi- cations. In the original planning for the project, fortifications were chosen as the focus of the research because of the wealth of information they would provide on the development and expansion of complex societies, and on the nature of interaction between polities and between environmental zones in the prehistoric Andes. Expansion of empires and interaction across space are two recurring themes in Peruvian archaeology, and it seemed likely that investigation of fortified sites in territories peripheral to and lying be- tween major central places would yield interesting results. After four seasons of field work we:feel that the utility of the research strategy has been confirmed. Our data and hence many of the tentative conclusions pre- sented here focus on issues of prehistoric militarism. But not surprisingly, the data have allowed us to draw inferences about subsistence strategies, demography, evolution of political systems, and inter-regional interaction in this very diverse zone. This report will emphasize the work done in the 1980 season. Results will be discussed from the survey in the Moche Valley, in the Viru Valley, and from the less intensive coastal survey north and south of that core area. A section on ceramic analysis will present a sequence for ‘the Otuzco sierra * ‘The research reported on here was made possible by grants from the Social Sciences and Humanities Research Council of Canada, and from Trent Univer- sity. Permission was granted for the study by the Institute Nacional de cultura, Lima. We wish to thank the students and assistants who partici- pated in the 1980 field season: Janet MacKenzie, Nicole St-Onge, Mark Borland, Eric Deeds, Alexis Deeds, Alfredo Melly Cava, Genaro Barr Argomedo, Manuel Tam Chan, and Marcelo Arroyo Rios.which supercedes information contained in earlier reports from the project. The cross-ties between the sierra sequence, ceramic collections from the middle valleys, and coastal styles will be discussed. The conclusions at ‘the end of the report draw on the data collected throughout the life of the project; while they must be viewed as tentative, we are confident that in outline they will stand the test of time and further analysis. THE MOCHE VALLEY The Moche Valley was a major focus of the Forts Project in 1980. Since good background information was available on the area (see Moseley and Day 1982 for an up-to-date bibliography), and since both principal investi- gators had previous experience there , survey was able to concentrate immediately on the section of the valley most in need of study. Fortification features are very rare in the lower section of the valley; only a few enigmatic "great walls" and two forts were encountered during years of survey by the Chan Chan-Moche Valley Project, while casual reconnaissance by the Forts Project 1377-1979 had shown a much higher frequency of fortifications in the middle valley. It is difficult to decide, at this point, whether the observed difference in fortification patterns between lower and middle valley areas is a result of differential preservation, different topographies, different military/defense strategies, or some combination of the three factors. It is common Andean practice to minimize labour investment in the construction of fortifications by selecting points on the landscape with maximum natural defenses; forts are commonly located on high steep hills with very difficult access. The Moche Valley in its lower reaches has relatively few such locations close to populated areas--fewer than in comparable valleys. But interestingly, several such points are ignored. Our attention was drawn to the middle valley section from elevations of 200 meters to circa 1200 meters above sea level (see Figures 2 and 3). ‘The middle valley (as we use the term) begins 18 km. from the Pacific Ocean where the peaks Of Cerro Oreja and Cerra Galindo bracket the Rio Moche and form a sort of gateway into the much narrower middle section of the valley; both of those hills are capped by Chim forts. In the middle valley, the floodplain is only 600 to 1500 meters wide, and is bracketed by very steep sided hills. ‘The river descends more quickly in this area; in the 18 km. lower valley stretch, the river drops only 200 meters, while in the equivalent 18 km. middle valley stretch the drop is 600 meters. This zone has been studied by both North Anerican and Peruvian archaeolgoists, but these investi- gators have tended to have fairly specific research interests. The middle valley has been examined with an eye to Early Man occupation, Formative ceremonial structures, and later prehistoric irrigation networks. No previous surveys or excavations had been oriented toward study of fortifications, and none had been broad enough in scope to approach the problem of coast~sierra interaction. In general, we found ample evidence for prehistoric utilization of the middle valley. Remains of domestic, ceremonial, and public architecture are clustered above the now-abandoned canals which follow the contours of thethe lower slopes of the hills. Wary settlements are located on very steep slopes which had been made suitable for habitation by construction of stone- faced terraces. In many sections of the middle valley habitation is nearly continuous along the valley margin and there is much evidence for re-use. — Preservation is a problem since many pre-Moche sites on steep slopes have been completed destroyed, probably by earthquakes. Similarly, the high frequency of Chimu sites on more level land--usually quebradas opening onto the valley floor--suggests that selective pressure on preservation is at work, and it is likely that many pre-Chim: sites in such locations have been washed away by the infrequent but severe flooding to which the area is prone. Clearly, settlement location above canals can reflect a shortage of arable land as well as defensive concerns, and this possibility was always kept in mind by project personnel. But the prevalence of defensive features (walls, ditches, slingstones) and the tendency of prehistoric settlements to fecupy the most inaccessible spots in the very broken terrain have convinced Je that during most phases of the ceramic era occupation of the valley, defensive needs were a very major factor in decisions about site location and layout. A total of 46 sites was surveyed by the project in the middle Moche Valley. Since the valley margin occupation is linear in nature and choice spots have several reoccupations, site designations were somewhat arbitrary, and many "sites" consist of several distinct sections which, after ceramic analysis, proved to have quite different occupational histories. The survey procedure involved written descriptions, sketches, and measurements of defensive features, and the relating of defensive features to non-defensive architecture at a site and to topographic Features of the vicinity. Excavations were carried out at Galindo and at Puente Serrano to clarify points of function, construction, and chronology of the particularly massive defensive walls at these two sites. Instrument maps were made of Loma del Shingo, Cruz Blanca, Siete Vueltas, parts of the Gaiindo defenses, and parts of Cerro Blanco del Leén. Ceramic collections were taken from carefully defined sections of all sites; all collections were analyzed in order to a) date occupation(s) of the site, b) assess evidence for contact with the sierra, and c) evaluate ceramic differences between the middle valley and the coast. Fortifications and Fortified Sites in the Middle valley No fortification features dating to the Early Horizon or earlier were encountered in the middle Moche Valley. At two sites, Puente Serrano and Cerro Petroglifo, poorly preserved mounds with conical adobes may date to the Early Horizon, but no other architecture could be associated with these features and it was clear that walls and other defensive construction postdated the mounds. Salinar sites were less conmon than in Viru to the south. The largest collection of Salinar ceramics is from excavated contexts at Puente Serrano, where a major occupation predates Middle Horizon defenses. It should be ted that very few Salinar sherds were found on the surface of the site, and analysis of only surface collections would have Jed to an underestimation of the ‘mportance of the Salinar occupation. This Salinar occupation occurs atboth the bottom and top of a tall steep hill, suggesting that defensive location was important, but no defensive architecture could be aseigned to the early occupation. While several other middle valley sites produced a very few diagnostic Salinar sherds (for example, steep-sloped neckless ollas with widely spaced burnishing tracks, very thin vessel walls, and broad white painted lines perpendicular to vessel mouths), there were no concentra- tions noted that required the presence of a major settlement. Huaca Menocucho, attributed to Salinar times by Kurt Brennan (1960:71), actually had few of these ‘classic! Salinar sherds. Cerro Arena, the major Salinar site in the lower Moche Valley, was briefly surveyed; no defensive features were noted, other than construction of part of the site on a relatively steep slope. There is more evidence of habitation and, presumably, defense during what is apparently a transitional Salinar-Gallinazo period. ‘The excavations at Puente Serrano allowed recognition of this time ceramically, when many Gallinazo traits involving plastic decoration and thicker-walled vessel forms are being added to the Salinar component. This transitional stage is picked up in abundance in the vicinity of Huaca Menocucho, directly across the valley at Cerro Cucho, and around the bend of the Rio Moche at Cerro Chile. The clustering of sites and the concentration on steep hills and quebradas lends a definitely defensive cast to the occupation, although no specific examples of military construction can be dated to this time. ‘There is a notable increase in number and size of full Gallinazo era sites. Sites include Frenteoreja, Quebrada las Terrazas, Puente Serrano, Cerro Huerfano, Santa Rosa Baja, and Santa Rosa-E1 Palmo, all on the north side of the river. Just inland from Cerro Oreja (which has a major Gallinazo occupation at its foot, largely covered up by later mate*ial) several contigu- ous quebradas have evidence of Gallinazo occupation, though few architectural remains. At the juncture of the Rio la Cuesta and Rio Moche, Cerro Menochucho has a Gallinazo occupation, as does Cumbray a few kilometers up the La Cuesta Valley. Upriver along the Moche, the site of Cruz Blanca proper and the adja~ cent ridges contain unmistakable evidence of a major and highly complex Gallinazo era occupation, matched across the river on a smaller scale by the contiguous hilltops of Cerro Chile, Cerro Dos de Mayo, and Cerro 1a Botija. Los Gentiles at the eastern Limit of our mid~valley survey also evidence Gallinazo occupation. th Most Gallinazo settlements in the middle Moche Valley are defensively Jocated on steep slopes and/or hilltops. Because of slumping, later construc- tion, and other adverse preservational factors, it is impossible to discuss artificial defenses in any detail. Wall segments were encountered at and near hilltops and running across quebradas between sites; on the basis of materials, measurements, and association, we assune that many of these wall remnants date to Gallinazo times but excavation would be required to confirm this assumption. At Cruz Blanca (Figure 4) we can quite securely date sophisticated defenses to the Gallinazo occupation. The site is constructed on a long and relatively steep ridge which rises from south to north to the La Cuesta-Moche dividing ridge northwest of the site (see Figure 3). The highest architecture at the site is of public nature, and contains poorly preserved terraces surmounted by large structures with foundations of large stones. Upslope from the highest architectural grouping, access from above is blocked by two deep trenches dug perpendicular to the ridge 2 to 3 meters into the rock. The slopes below the site center were once covered by smallnba stone-faced terraces and domestic structures, but these have been almost com- pletely destroyed. This Gallinazo equivalent occupation of the middle valley is obviously an intensive one. We argue below that one factor leading to the buildup cf both population and fortification was an increase in trade and communication with the sierra; ceramics from Cajamarca and the Callej6n de Huaylas as well as fron the Carabamba Plateau and other adjacent sierra regions are found at Cruz Blanca and at some small sites in the middle valley, in contexts which strongly favour the Gallinazo date. These sierra sherds are not found in either excavated op surface contexts at Puente Serrano, which had a Gallinazo occupa- tion, and they are not found at all in the lower valley to the best of our knowledge. Since no definite Moche I-II occupation was noted in the surface collections from the middle Moche Valley, it is possible that this Gallinazo equivalent occupation is in its later part contemporaneous with and forms the local chronological equivalent to the lower valley's Moche I-II, In this case, the middle valley would be more similar ceramically to the Viru Valley during the (approximate) years 1-200 A.D. Several areas of major Moche occupation were found in the middle valley, and minor Moche presence at most other sites was noted. No Moche I-IT markers were found at any mid-valley sites, including cemeteries. Phase III-IV markers were quite comnon, and Moche V was present but there are pro- blens with the identification of this phase. Essentially we used castellated florero rims and the Moche V conical spout and angular stirrup as markers. None of the other ceramic characteristics discussed by Bawden (2977) or Donnan (1976) was found in either excavated or surface collections. Santa Rosa Baja, E1 Palmo, and Santa Rosa~El Palmo all have substan- tial Moche ITI-IV occupation. Santa Rosa-E1 Palmo and El Palno are unfortified settlements of moderate size; Santa Rosa Baja is something of an anomaly, in that defenses and ceremonial architecture are associated in ways which are unparalleled elsewhere in the middle valley. On the opposite side of the _ valley an incomplete survey did not pick up any significant Moche III-IV occupation. A major cluster of Moche habitation is located west of the juncture of the Rio la Cuesta with the Rio Moche. An essentially continuous occupation at Cerro Menocucho, Panteén, and Pedregal Bajo is an expansion of the earlier Gallinazo presence at Menocucho. Cemeteries and domestic structures in this area can be dated Moche III-IV by surface evidence, but fortifications appear to be later. Pedregal Bajo produced Moche V ceramics, as did Cumbray a short distance up the Rio la Cuesta. A third Moche III-IV cluster occurs around the modern town of Poroto. Huacamochal (unfortified settlement), Los Gentiles and Siete Vueltas (small hilltop forts) and Cerro Poroto (terminus of a ridge-route from the Carabanba Plateau) all have Noche III-IV occupation. This occupation is far more minor than that further downriver, however. There are some madi ifts during the Middle Horizon. The Poroto sites are all abandoned but the site of Cruz Blanca downriver is reoccupied.New structures and defensive walls are constructed at the site, but the overal) extent of usage is less than in Gallinazo times, and the hilltop forts above the site are not re-used. Across the valley, strategic hilltops at La Botija and Cerro Chile are occupied, probably for defensive/strategic purposes. At the confluence of the La Cuesta and Moche, Pedregal Bajo continues to be occupied in Noche V but the nature of that occupation is not entirely clear, and Cunbray is reoccupied a short distance up the La Cuesta. The largest single Noche V site in the Moche Valley is Galindo, located at the point where the middle valley begins. This urban site is built on the lower flanks of a high steep hill and in a quebrada which drains the area behind the hill. ‘The site must have had a population of several thousand, and it contains some impressive public architecture (Conrad 1974). Garth Bawden (1977, 1982) conducted excavations in many parts of the site and produced a map and descriptjon of the site as a whole. Me returned in 1980 to a feature of the site which had not been excavated--the 1.5 km. long wall which skirts the bottom of Cerro Galindo, crosses the major quebrada, and jeer terminates on a low hill west of the site. The wall had been interpretted as 2 social barrier. Excavations proved the wall to be defensive, at least oa with in its early stages. It was parapetted for most of its length, and in the ante sections where the wall top is preserved a thick carpet of slingstones caps Gabe the top. The slingstones can be distinguished be the high degree of uniformity of weight and sphericity displayed; they were brought up from the quebrada below the wall. Access through the wall is restricted and protected. The wall is truly massive, At one section where excavations were carried out behind, in front, and on top of the wall, the base was 4.5 meters thick, the top was 3.5 meters thick, and the outside preserved height was 4 meters. The wall was maintained and reconstructed over a long time. Gateways were blocked and new ones constructed from time to time. Where the wall crossed the quebrada, it had been washed out at least three times and rebuilt. We may be able to deduce an easing of tensions with time, as by the end of the occu- pation of Galindo much of the population lived outside the defensive wall. ‘The Moche V occupation continued around the southern slope of Cerro Galindo at Frenteoreja, but there is no evidence of Moche V use of the top of Cerro Galindo. Puente Serrano is a major Moche V site, as heavily fortified as Galindo. A massive parapetted wall seals off a very steep quebrada; the site was occupied both before and after Moche V, but the major defenses date to this period. The other Moche V site in this part of the middle valley is directly across the valley at Cerro Blanco del Le6n, Here an isolated hilltop with poorly preserved walls is bracketed by quebradas which provide more or less direct access to the Viru Valley. The Moche V evidence comes chiefly from cemeteries and there are no extant fortifications of Moche V date. Thus the mid-valley Moche V occupation is clustered at three points which control communication: Cruz Blanca-La Botija controlling the upper middle Moche Valley; the several sites at the confluence of the La Cuesta and Moche Rivers; and Puente Serrano-Cerro Blanco del Leén overlooking the pass to Vir. These sites are clearly outranked by Galindo itself, which controls the interface between coast and middle valley. The Chimu occupation of the mi che Valley ig intensive and well- preserved. I¢ is quite rare_in the middle valley to walk for more than 10 ‘meters without encountering a Chiu sherd. There are several different kinds of Chima sites, occuring in different sorts of locations, but it is ingeneral unusual for a Chimu site in the middle valley to completely lack defensive features. One of the more common sorts of site is the hilltop fort.° These forts are not particularly elaborate. A space at the top of a hill is cleared and a protective wall and two or three simple structures may be built, Salient observation points may be selected and improved a short distance below the summit, The defenses rely not on sophistication of con- struction but on their position. It is generally at least a one-hour scramble up to a high Chimu hilltop fort; this climb is undertaken in full view of the summit, and on at least two hills (Cerro Galindo and Cerro Oreja) small defensive terraces and walls give defenders locations from which to harass attackers. Stone, abundant on these hills, were probably the weapon of choice. It is possible that these strategically placed posts were intended more as observation points than as garrisons. In the case of the higher hills, the Chinu fort is usually the first site built on the location; the pattern of the earlier occupants of the middle valley had been to construct settle- ments on moderate-sized hills and to crown the hilltop with a fort, but not to occupy the really inaccessible hills. The Chimu are the first to separate settlements and defenses as they construct special purpose fortifications. The earliest two hilltop forts (and the earliest Chimu sites encountered) are the twin forts of Cerro Oreja and Cerro Galindo, which overlook and control the valley neck where the transition is made from wide coastal flood plain to narrow middle valley. Cerro Oreja was built first, but the two were probably in use simultaneously. Puente Serrano is reoccupied later, but there is no important rebuilding or improvement of the earlier defenses. Santa Rosa Baja is reoccupied and some defenses may be of Chimu date. Two Chima forts control the juncture of the Rio 1a Cuesta and the Rio Moche--one on Cerro Jesus Warfa and one on Cerro Pedregal. In both cases, the Chimu fort is the First on the location, and the hills are very difficult of access. cunbray, Riobar, Dos de Mayo, and Cerro Chile all represent use of hilltops for defensive purposes, with varying degrees of associated settle- ment. Undefended settlements in this lower part of the middle valley are generally fairly late (cambered rims and angled plates), and include Cerro Rivero, Cerro Blanco, Cerro las Terrazas, and Cerro Petroglifo. Settlements below ‘Chim forts include Cerro Cucho, Pedregal, Pedregal Bajo, Cerro Panteén, Cerro Menocucho, Jesus Marfa Village, and scattered structures on the lower slopes of Cerro Chile and Cerro Dos de Mayo. The area surrounding Poroto (see Figure 3) is especially interesting during the Late Intermediate Period. Cerro Poroto itself is the lower part of a ridge which descends quite directly from the Carabamba Plateau 12 km. to the southeast horizontal distance but 2200 meters higher. On the lower part of the ridge, a series of seven clusters of architecture were found, linearly arranged. The structures are small, scattered, and undefended, and probably served travellers ascending or descending the ridge route. The ridge though {a bracketed on the east and west by formal forte which wane used by the Chim. |“ Directly across the river from Cerro Foroto is the large walled town of Loma del Shingo (Figure 23). The site is built ona steep-sided ridge, and defended further by an encircling wall and trenches on the upslope end of the site. The layout is highly regular and obviously planned, although the units are probably domestic. Construction and peopling of the site was probably an imperial Chimu project and the location across from one of the main routes into the sierra cannot be coincidental. Two kilometers further west on the same side of the river as Loma del Shingo, Pefla Larco and Puente Larco com- bine domestic and public features, including storage facilities. It appearsthat the sites in the Poroto area are specializing in contact and probably exchange with the neighbouring sierra. The proportion of sierra ceramics on these sites is much higher than on mid-valley Chiu sites downriver. No evidence of Inca conquest, administration, or habitation was found anywhere in the middle valley of the Rfo Moche. SURVEY IN THE VIRU VALLEY The purpose of the survey in Viru was to collect a sample of forti- fied sites from the lower and middle valley zones of a northern drainage con- parable to the Moche Valley. It was expected that expanding the sample in this way would allow us to gauge the variability present in the developnent of fortifications through time in different areas of northern Peru. In addition, we wished to know whether another valley adjacent to the Otuzco sierva would contain ceramic evidence for coast-sierra interaction like that found in the middle Moche Valley. Viru was of course a logical choice for this ancilliary study. The work of the Viru Valley Project in 1946 had been well-published and a great deal of data pertaining to the settlement pattern study (Willey 1953) and the ceramic dating of sites (Ford 1949) was readily available. Of the 316 sites reported by Willey, 32 were revisited and several others were subsumed by new site designations (see Figure 5). The majority of the sites were located along the edges of the irrigated lower valley or on the slopes over- looking the Rio Viru in the narrow Huacaponge area. Hilltop locations were common and site settings in general were similar to those in the Moche Valley. Many of these revisited sites had been classed by Willey as "castillo forti- fication complexes", "hilltop redoubts", "hilltop fortifications", or as "pyramid-dwelling construction complexes with some fortification features." In all these cases, we were in agreement with Willey's assignation of a fortification or defensive role to the sites. Other sites however which had been classed as irregular agglutinated villages or terrace groups frequently turned out to have unmistakable defenses associated with them. These defenses, usually walls controlling access to sites, were neither described nor included on site maps in the 1953 publication. This is understandable since in many cases these defensive walls are incompletely preserved and/or located some distance fron the architecture forming the core of the site. This is particularly true in the mid-valley area where the Rio Viru makes its eastward sweep (see Figure 5). Habitation sites both north and south of the river are interconnected and protected by a complex wall system whose major epochs of construction and use were apparently during the Salinar and Chimu occupations of the valley. Willey's survey of the valley was remarkably complete. In the part of the valley covered by the 1946 survey, we found very few unrecorded sites. Most of the 42 new sites defined by the Forts Project were located outside the Limits of the earlier survey, especially in the upper Salinas, Viru, and Huacapongo drainages. Generally, fortification patterns in the Moche and Viru valleys displayedmany similarities. In both valleys there are comparatively few fortifications in the lower valley areas. The Viru Valley does have more coastal fortifi- cationis than the Moche Valley; Bitin, Bitfn Bajo, Chuquil, Chuquil Bajo, Cerro las Pifias, Cerro Sausalito, and other sites both large and small (see Figure 4 and key) all have defensive characteristics. In part, this higher frequency of coastal defenses may reflect topographic differences between the two valleys. A hill like Cerro Bitfn, for example, is an ideal location for a fort; its isolated position in the lower valley gives it an excellent view for miles around. Several of the Viru lower valley sites (including Bitin) have major Salinar components. The Salinar occupation in Viru, as preserved, is much larger and more important than in Moche, and fortifications and places of refuge are common in Viru during this period. But in Viru, as in Moche, the majority of fortified sites occurs in the middle valley. Im Viru, Cerro Napo and Cerro Tomoval form the "neck" of the valley, marking the point where the wide coastal plain becomes constricted. Upriver in the mid-valley zone, the floodplain is narrow and steep hills lie close to the rivers and quebradas. The Viru mid-valley zone resembles that of Noche in that evidence of fortification (walls, walled terraces, and forts) are found on almost every hill. But in Viru, we find that the occupations of the various defensive locations are nearly continuous. For example, a surface collection from Cerro Vinsos Bajo (#31 on Figure 5) includes sherds of Puerto Moorin, Gallinazo, Huancaco, and Chimu date, as well as sherds from the Carabamba Plateau in the adjacent sierra. Because of this frequent reoccupation of sites, survey data alone do not allow a detailed phase by Phase discussion of the evolution of fortifications through time in the valley. Another similarity between the Moche and Viru Valleys is found in the /iisiribation patterns of sierra ceranics. Sierra sherds are very infrequent in the coastal part of the valley, and more common in the mid-valley zone. In Viru the kaolin sherds which evidence contact with distant parts of the north and north-central Peruvian sierra were present but infrequent. The majority of exotic sherds come from the Carabamba Plateau, only 18 km. NE of Los Tres Condores (#61 on Figure 5). It is clear that there is a great deal of interaction occuring between the chaupiyungas settlenents of Viru and the serranos of Carabanba. According to the ceramic evidence, this interaction is strongest in the Gallinazo and Chimu eras. It is interesting that the only site which produced local copies of Incaic pottery in our survey of the valley is Cerro Lechuza, a site located well up the Rio Carabamba, distant from the coastal part of the valley. In general, fortification patterns in Viru are similar to those in Moche in the following ways. 1. There are no Early Horizon fortifications. 2, The intensity of fortifications in general is highest in the middle valley. 3, The most common defenses in earlier time periods are places of refuge and single site defenses. Strategic fortifications protecting land and ~ boundaries are more important in later times. 4. Improvement of defensive military capabilities through time results iess from technological improvement than from better location, layout, and coordination of defense structures. Presumably a sinilar increase in the ability to muster, train, supply, and utilize professional and part-time military personnel would also have occurred, but this kind of change is extremely difficult to read in the archaeological record.-10- COASTAL SURVEY An informal survey of forts located in the coastal sections of valleys north and south of Moche-Viru was carried out toward the end of 1980. The main focus of the survey was Chimu fortifications, as an attempt was made to track the expansion of Chimu territory in the Late Intermediate Period; some earlier sites were visited however. We should stress that our coverage of the Santa, Chao, Chicama, and Jequetepeque Valleys is by no means complete, and is only meant to provide auxilliary information on coastal fortifications. The fortification feature with earliest dating was picked up at Salinas de Santa, a short distance north of the Santa Valley. Here, a few hundred meters north of the western terminus of the Great Wall of Santa (studied, along with other prehistoric fortifications of the Santa Valley by David Wilson of University of Michigan in 1980), a small preceramic site is located on an old beachline. The site is of probable Cotton Preceramic date (Moseley, personal communication). On two sides of the area of densest refuse, lines of stone piles run inland from the beach. The space under discussion Forms a trapezoid, with the beach being the top, the stone lines the non-parallel sides of the figure, and the longer bottom left unenclosed. The stone piles are unusual for a variety of veasons, and are tentatively identified as slingstone piles. ‘The actual piles are apparently small (although no excavation was done to check the size of the piles), consisting of 2 or 3 dozen stones each. Not wanting to disturb the stones, we took no measurements on ranges of size and weight, but impressionistically they were the size of medium- to large-sized hens' eggs, but less rounded, The piles are spaced 2 to 3 meters apart, running for a good 30 meters, one line on the flat and one line up a ridge. ‘There is no source for these stones in the immediate vicinity. The two lines of stones i ith_any sort of wall or structure. If they are a defensive feature, it is odd that ‘The inland side of the site was not provided with a line of stone piles. Other interpretations of the stones might be possible, but in our experience clusters of stones of this size and shape are invariably associated with defense. Two forts were examined in the Chao Valley. One of these, Cerro coronado, was described in an earlier report (Topic and Topic 1978). Briefly, it is a rather small double-peaked hill beside the Pan American Highway which overlooks the coastal part of the valley and controls coastwise movement (see Figure 5). The concentric walls comprise a fort on one peak, while domestic terraces (many unfinished) blanket the lower slopes and the adjacent peak. The ceramics are Phase 2 Chimu, relatively early in the Chima sequence and roughly contemporaneous with early fortified Chimu sites in the middle Moche Valley. Ten kilometers to the NE, Cerro de 1a Cruz is a sizeable town built on a hill protected by two encircling walls and a trench (see air photo, Kosok 1965: 185). There are_slingstones on the benches of ‘he parapetted enclasing wall. There are also, more unexpectedly, piles of -SLingstonee(outsideShe defenses, but a careful search failed to provide Final evidence ‘ingstones being hurled from either line of defense. The surface ceramics within the site are a mixture of local and Chimu styles. It is quite possible that at this site we see the Chimu takeover of an important local refuge, probably through intimidation rather than conflict.att. In the Chicama Valley north of the core survey zone, a cursory survey of several hilltop sites was carried out to see whether the general sorts of patterns uncovered in Moche and Viru would be repeated. In fact, they appear to be. While sites in the lover valley are sometimes located on available hills and rock spurs, these may be unfortified. Piedra Parada is an example of this situation; it is a Chimu settlement built on a rocky ridge overlooking Chiquitoy Viejo, only 9 km. from the ocean. A wall bisects the ridge, but Serves no apparent defensive function. On the northern edge of the valley another Chimu site, Cerro Pan de Azucar (Figure 6) overlooks an algorrobo "forest". Its position against # hill and the presence of walls suggest defense was a factor in site location, but the overall architectural patterns indicate more of an elite and administrative presence. Some 32 km. inland from the ocean, however, where the valley hegins to narrow and rocky hills overlook the flsod plain, unambiguous fortifications of Chimu and pre-Chimu date are frequent (Figure 7). Piedra Molina, a low rock spur on the south side of the river at this narrowing point, has successive Salinar, Gallinazo, Moche, and Chimu occupations. The ceramic collections also praduced exotic sherds; jars of 2 low grade greyish kaolin paste were painted with finely done spirals and parallel lines executed in purplish-red (Figure 8). The only other occurence of such sherds was in viru, at Cerro Pongo (V-212). ‘The Muralla 1a Monica and Muralla Sausal (see Figure 6) are two wall segments, 1 and 2 km. long respectively, whose construction probably dates to Moche III-IV. Both wrminate at Cerro Sausal and may control access to/from the quebrada behind. They are not associated with distinct fortifications, and probably are not strictly defensive in nature. Cerro Facalf is the most interesting fortjfied site of Chim date in ‘the Chicama Valley. It is located on the northern margin of the river. The site is an enormous place of refuge, formed by the building of up to 10 concentric stone #alls around a large oblong hill with very steep and rocky slopes. There are few structures inside the walls, and no structures on the summit. ‘This rather enigmatic site looks to us clearly defensive, and it seems to be completed but never used. Two Late Intermeidate Period sites were studied in the Jequetepeque Valley (Figure 9); they fulfill expectations of what LIP fortifications should look like during the Chimu expansion. Cerro Talanbo is a very large site, with several walls enclosing a hill and a variety of architecture within the walls, including elite, public, and domestic architecture (see Kosok 1965: plate on pp. 124-125). Several transects made outside the main wall indicate that a battle probably took place here; slingstones are found along the top of the outermost wall, but also on the ground outside the wall. Measurements indicated that the frequencies of the slingstones (identifiable because of their uniform size and greater than normal sphericity) increase up to about 40 meters out from the wall, then decrease until they stop at &0 meters from the wall. Many of the slingstones are broken. The ceramics are not imperial Chima, but a local early Late Intermediate Period style, based on preliminary analysis. The second site examined IS an imperial Chimu fort, built on a hill overlooking the Chimu administrative center of Farfin. This fort, Cerro Faclo,-12- is a specialized construction, probably built to protect Farffn. An elongated oval area at the top of the hill is enclosed by a defensive wall. Various spurs that would allow access up to the hilltop are blocked by short wall segments. There is a road climbing up to the sumit, bat a control struc- ture blocks access through the wall. Small piles of slingstones are found in various parts of the fort. Ceramic evidence agrees with lack of rebuilding and scarcity of refuse in suggesting a short occupation of the site. It would seem that after pacification of the area, the fortification was abandoned. CERAMEC ANALYSIS In 1980, analysis of ceramic collected from sites surveyed by the Forts Project was a major thrust of the research, During the year, current collections were analyzed and most collections from previous seasons were re-examined, The end result of the analysis is relatively secure dating of over 90% of the sites surveyed by the project, and tantalizing but fragmentary data on the nature of exchange and interaction in prehistoric northern Peru. A brief review of the nature, aims, and problems of the ceramic analysis over the life of the project is in order. During the first three seasons of the project, the focus of the research was on the sierra part of the survey zone, in the vicinity of Otuzco. Some 135 sites were identified and studied in the area, and collections of surface ceramics were taken from most of these sites. Construction of a ceramic sequence was difficult in the extreme, since the collections were: 1. surface collected, with no stratigraphic information; 2. lacking alnost entirely in trade ceramics from known sequence, with the exception of small amounts of Chimu pottery at a handful of sites; 3. in many cases comprised of very weathered and fragmentary pieces; and 4. heavily weighted toward undecorated utilitarian wares. Overall, 12% of ‘the 17,000 sherds from the sierra part of the survey zone were decorated, and the vast majority of this decoration was exceedingly simple--e.g. a single red painted line. In 1977, seriation of collections and use of the limited cross-dating information available allowed the setting up of a tentative ceramic chronology which was refined in 1978 and 1979. In 1980, with the shift in emphasis to the coastal part of the survey zone, one of the main aims of the ceramic analysis was to verify the ceramic chronology developed for the Otuzco area. Since we had found very few coastal ceramics in the sierra, and since our experience with coastal sites in 1969-1973 had indicated an almost complete absence of sierra ceramics on the coast, We were not especially hopeful of finding good data for cross-checking the coastal and sierra sequences. Fortunately however, in the middle valleys of the Moche and Viru Rivers dur- ing 1980 we surveyed a mmber of fortified sites which contained quantities of both sierra and coastal ceramics. Anticipating the Conclusions section of this paper, we can state without doubt that the middle valleys of these rivers were occupied by intermittantly independent groups which, at several points in the ceramic part of the sequence, were heavily involved in faci- Litating and, perhaps, regulating coast-sierra exchange.-13- The result of this extrenely useful cross-dating data was to confirm the order in which sierra ceramic phases had been placed, but to push the earlier phases back further in time than we had reported earlier (Topic and Topic 1978, 1979). We have defined four ceramic periods for the Otuzco area, from ‘the early Early Intermediate Period to the end of the Late Intermediate Period. In addition, Inca presence in the area was detected, but does not correlate with any changes in ceramic technology or style. This four-part subdivision of the sequence is not particularly fine, we know. But given the fact that the sierra ceramic collections are from the surface, mostly undecorated, and largely lacking in useful (as of this writing) cross-ties to other sequences, we feel that it is the best that can be done until other lines of analysis have been tried. ‘Two other problems with the sierra ceramic sequence might be of inter- est to some readers. First, the Otuzco area is ceramically a very conserva~ tive one throughout its prehistory. Forms, clays, and firing conditions are extremely persistent, especially in smaller sites. "Elite" wares display a greater Frequency of change, but are found mainly in large sites, and-frequen- cies fluctuate fron period to period. Secondly, our survey zone was so large as to encompass distinct ceramic "territories" whose boundaries fluctuate and whose styles do not change at the same rates (see Figure 10 for a rough outline of the zones discussed below). Overall, we have chosen to be conserva- tive in the framing of the ceramic sequence. Me will briefly review the four sierra phases and their geographical distributions here. The Sierra Sequence Phase 2 is tentatively dated to the early part of the Early Intermediate Period. Tt is characterized by sand tempered pastes in the 5 to 6 mm. thick- fess range, high percentages of neckless ollas (over 50%) in collections, oxidizing firing atmospheres, and by a paucity of decoration. At some sites relatively Fine thin-walled hemispherical bowls and incurving bowls are conmon, often red-slipped. Jars with thin convex flaring necks occur, infrequently with red lines painted at the vessel lip. At two sites in the eastern part of the survey zone, colander forms (open heimispherical bowls 6-8 mm. thick) occur with angular holes; this form is also common at Cerre Campana East near Huamachuco, a site dated to the latter part of the Early Horizon (Thatcher 1972; Topic and Topic 1982). Kaolin is absent at these early sites, Phase 1 48 recognized at 1 site in the Otuzco area, two sites on the Carabamba Plateau, and 3 sites further east in the Motil area, Population densities must have been quite low in Phase 1, and there is a decided preference for high elevations. Phase 2 is equivalent in age to Gallinazo-Moche on the coast (c. 200 B.C. to A.D. 500) and is represented by far more sites than either Phase 1 or Phase 3; with further analysis of the ceramics and refinement of cross- dating to the coast and to Huamachuco and other sierra points, it will be possible to subdivide the phase. At present, two distinct styles ate assigned to this Phase, the Northern and the Southern. The Northern style is charac- terized by a very tight clustering of form, paste, and decoration. The most common form is a round-bottomed jar with a flaring neck with a pronounced convex curl (see Figure 11), It is common for the lip edge of these jars to be grooved, indented, or made bulbous. Rims of this general shape occur on both large and medium sized jars. The only decoration occuring (infrequently) is plastic. An appliqued filet, usually punctated with circular impressions,n14- can oceur (see Figure 12). In a handful of cases, the filet resembles an animal. This mode of decoration has coastal parallels in the Moche and Viru Valleys in Gallinazo and Moche times (Bennett 1950; Topic 1977). Paste colaur, both surfaces and cores, ranges from orange to blue-ish grey de- pending on the nature of firing, but the paste is always hard and tempered with small bits of dark rock; surfaces are wiped and unslipped. This Northern style is the predominant one on the Noche-Chicama Divide, comprising up to 80% of the collections made by MacKenzie (1980). This style is not confined to the Divides it is found in lower quan~ tities in the Otuzco Basin, on sites in the Mache area, and in Motil. It is closely related to a style found at Marcahuamachuco and neighbouring sites (based on preliminary analysis of collections made by ws in 1981), where the same clustering of shape/form attributes (but not decoration) occurs. In Huanachuco, the type grades into softer and paler (e.g. kaolin content) analogues. The style does not occur in the middle valleys of either the Moche or Viru Valleys. In the move southern and eastern portions of the sierra survey zone, Phase 2 ceramics are less uniform. There are steady frequencies of utilitarian wares, usually undecorated flaring jars in browns and reddish browns; paste thinness, texture, and inclusions vary from one zone to another. There is some carry-over of angular colander, but they are rare. Buff-coloured paste is a hallmark of the Southern style. Pastes are quite Light but not within the "kaolin" range and sherds often have a thin dark grey unoxidized core. The commonest forms are jars with everted rins (see Figure 13). Frequently there is a groove or indentation on the ip exterior approximately 1 om. below the lip edge (see Figure 13A). _Approxi- mately 90% of these jars are decorated with red painted lines. A 2-4 cm. wide red band may occur at the lip interior or extenior, or the entire inside or outside of the neck may be painted red (usually not both). & red line or band may be placed at the rim join, in combination with some band or line at the lip. Rarely, red Jines are painted on the neck interior perpendicular to a red band at the vessel lip. This buff jar is very distinctive and has a wide range of occurence. It is found in major sites on the Carabamba Plateau like Suicha, Mollepuquia, and Paredones. In the Motil area, it is conmon at several sites which also have the Northern style, and at one site (Campana Chica) which has Huamachuco trade wares. In the Moche area, intermediate between Carabamba and Motil, a similar buff jar occurs but it is usually unpainted and is exceptionally hard (see Figure 198 for typical rim shape); this variant occurs at Pefla Blanca and several nearby sites, as well as at Sango in the Motil area. The classic buff jar occurs in the middle Moche Valley at Cruz Blanca, Santa Rosa Baja, Huaca Menocucho, Dos de Mayo, Cerro Blanco del Le6n, Pefia Larco, and Puente Larca. Throughout the range of the buff jars they occur in association with small Fine bowls. These are generally simple flattened hemispherical forms (Figure 16, 17, 18) although rarely a slight or pronounced angle may occur on the side (Figure 144, 178). Bases are always rounded; tripods do not occur and ring bases are exceptionally rare and probably later. Lips are generally simple (Figure 16 4,B,C) although lip shaping may occur (Figure 14 B,C,D,E,G). Paste thickness is 4-5 mm and vessel diameter is generally between 16 and 24 om,-18- Within these shape ranges, to basic pastes occur--fine brown and kaolin. The fine brown bowls are most commonly decorated in red slip or red painted lines, although a red-white-orange sub-type occurs in the middle Moche and Viru Valleys and on the Carabanba Plateau (discussed below). The brown bowls occur at nearly all Phase 2 sites. The kaolin bowls are less common, and tend to be most frequent at the larger sites. They show a wide range of variation in decoration, almost invariably on vessel exteriors. Red-on-white is the most common color com- bination; simple lines and panels around lip interiors and exteriors are the most common designs, followed by simple lattice (cross-hatching) designs in a band along vessel exteriors. Other simple geometric designs occur (Figure 16 C for example). Other colour combinations include black/orange (Figure 16A and B), brown/white, orange/white, and brown/orange (Figure 178). Three colour combinations include red and brown/white (Figure 15C), black and orange/cream, brown and white/grey (Figure 17A), and red and orange/white. No complete vessels were recovered, and most sherds displayed only small portions of a design. Many of the examples in Figure 16 through 18 are taken from middle Moche Valley contexts, primarily because of the better preservation of painted designs on sherds in the drier context. It mst be stressed though that there is great similarity between kaolin bowls of the sierra Phase 2 and Gallinazo(Moche I-II equivalent) sites in the middle Moche and Virw Valleys. Two types of plastic decoration occur on kaolin bowls. One is nicking or scalloping of bowl lips, which ranges from slight fingernail nicas to heavy pie-crust-like scallops, to actual castellation. Generally, this occurs on unpainted kaolin bowls, but sometimes it co-occurs with red paint. Nicking and scalloping never occur in combination with other paint colors. The second type of plastic decoration consists of a single incised line running parallel to and about 1 on. below the bowl lip on the exterior (see Figure 18). ‘he line is always single, ané never occurs in combination with painting. Vessels of this style often have an orange tinge to them which in most cases is not a slip but a result of firing and/or clay characteristics. This type of bowl is found throughout the survey zone, and appears to persist through Phase 3. It has proved very difficult to sort out the sources of these various kaolin bowls. Some are apparently local, others are local copies of imports, and sone are clearly imports. In another place (Topic and Topic n.d.) we use the kaolin styles to trace middle EIP communication networks in the north sierra, citing similarities(and in some cases replication of) kaolin vessels from Cajamarca and from the Callején de Huaylas. ‘These relationships are still being sorted out, and preliminary results indicate that very complex interactions are taking place in the north sierra and adjacent areas between c. 200 B.C, and A.D. 500. For the present, we wish to emphasize two aspects of the kaolin bowls: 1. the different styies ave found in the sierra (especially on the Carabanba Plateau) and in the middie valleys; these kinds of bowls are apparently not reported fron the coast. Phase 3 is tentatively correlated with the Middle Horizon, although we lack convincing cross-dating; no Huari or Poche V sherds were found in the Otuzco area, and only one Cajamarca Cursive sherd--from an overwhelmingly LIP (Phase 4) context. Painted kaolin bowls drop out although fine brown bowls~16- and incised kaolin bowls continue. Angular-holed colanders are replaced by colanders with round holes. Jars with concave curves begin to appear, while the flaring jars of Phase 2 continue. On the Moche-Chicama Divide the distinc- tive Northern style is diluted; the jar shapes persist, but in a sandier brown paste and in association with a variety of new vessel shapes indicated more contact with Motil, Moche, and the Otuzco Basin. Sites in the Otuzco Basin are eclectic; even small Villages echo the Northern style shapes, have buff paste reminiscent of the earlier Southern style, but are anticipating some LIP shapes. Overall, Phase 3 is the most poorly defined of the four phases. It seems transitional from the better defined Phase 2 to Phase 4, and displays considerable variability from one site to another. Phase 3 marks the aban- donment of many large sites, espetially on the Carabanba Plateau, and a trend to abandonment of high altitude sites on the Moche-Chicama Divide and in Motil, with a corresponding increase in populations at lower elevations in zones such as the Otuzco Basin. There is no convincing evidence of interaction with the middle valleys of Moche and Viru. Clearly, this is a time of political and economic reorganisation in this part of the northern sierra. Phase 4 ceramics show a high degree of uniformity throughout the sierra survey zone. Fine wares have alnost disappeared; kaolin is extremely scarce and only a few fine bowls, black or brown, occur. Three forms are very common--large jars with flaring necks (Figure 19A, Figure 20A and B), jars with incurving necks (Figure 19C and D), and bowls with round colander cut-outs. Pastes and surfaces are brown, The colanders have very rough exteriors but smoothed interiors. Temper can include sand, crushed rock, and crushed sherds. Gold mica is a frequent inclusion in Phase 4 pottery fron the Carabamba Plateau but not from other areas. The incurving neck jars occasionally have lugs at the lip exteriors; usually they are oval, with a nick or depression across the short axis. These split-oval lugs are always positioned with the long axis parallel to the vessel lip. Less common is a conical lug. The everted jars are frequently decorated with a single painted band at the lip interior (Figure 20A and B). This is usually a thick red paint. Less common is a thick orange band. Very rare is black and orange. ‘The paint, whatever the color, tends to be unevenly applied and to have an indistinct "smeary" lower border. Neckless ollas occur but are not common. A Chimu-Inea spout and a probable aryballoid rim were found at Rogoday, on the Moche-Chicama Divide in association with a terrace complex covering several hillsides. There were no differences between the general Rogoday colkction and general Phase 4 collections from Carpaico downslope. COAST-SIERRA TIES We have mentioned in the previous sections the scarcity of coastal ceramics in the sierra of Otuzco, and the lack of sierra sherds on the coast. A total of about 12 Chimu sherds was found in the Otuzco area by the project local collectors have Chimu vessels in their collections, many of which are purported to come from tombs in the vicinity; none of these claims could be verified. No recognizable Moche, Gallinazo, Salinar, or Cupisnique ceramics occurrs in the sierra part of the survey zone. Similar, coastal Moche-17~ Vaticy sites at which we had worked in previous years did not produce ceramics like those from Otuzco. But in the middle valleys of Moche and Viru sierra sherds do occur, at times in quite high frequencies. Ceramics of Phase 2 (Gallinazo-Moche T-II equivalent) and Phase 4(LIP) are especially well represented, In some cases we can only state that sierra sherds are present at a site. In other cases, they are restricted to certain sections of a site. On other cases, sierra sherds are the dominant element in the ceramic collections. During Phase 2 the Southern style is well represented in the middle Moche Valley. Buff jars (see Figure 13A and B) with red painted lines occur at several sites. Some of these are located close to Poroto, near the terminus of the ridge-route (Co. Poroto, Huacamochal, Pefia Larco, Puente Larco, Cruz Blanca, Dos de Mayo, La Botija). Two other sites, Santa Rosa Baja and Cumbray, have in gemral a higher than usual proportion of sierra sherds, especially kaolin, Co. Blanco del Le6n at the entrance to the middle elevation pass to Viru and Huaca Menocucho (a site of ceremonial importance for many centuries) have buff pastes. Many of these sites have in addition two kinds of short-necked ollas (Figure 13C and D) which also occur at Paredones and other sites on the Carabanba Plateau. While the sierra forms are usually in buff paste, the mid-valley forms are usually in a dark brown paste with heavy sand temper. In fact, the middle valley during Phase 2, especially upriver of the La Cuesta-Moche confluence, has high frequencies of ollas, neckless ollas, and bowls in sandy dark brown paste. Though the forms and decoration are similar to the classic Galiinazo repetoire described by Bennett (1950), it is quite clear that at sites like La Botija, Co. Blanco del Leén, Dos de Mayo, Co. Chile, and Cumbray we are dealing with a distinct ceramic traditons this ceramic tradition was present at Panteén, but absent downvalley at Puente Serrano. Associated with the mid-valley imported buff and local dark brown pastes is a distinctive Red/White/Orange painted ware; this type of decoration is usually executed on the dark brown pastes, often (but not invariably) on bowls, and consists of simple geometric designs and bands (see Figures 20 and 21). Bennett (1950:96) in his study of Gallinazo sites encountered a jar of this type at the Gallinazo Group, and dubbed the type Castillo R/W/0. Onty one example is mentioned, and the provenience is given as "intrusive". All of the 13 sites he excavated were in the lower Viru Valley, all quite close to the Gallinazo Group, so it is not surprising that the middle valley type did not Figure more prominently in his classification of collections. Ford (1949) does not mention the type in the text; he was however impressed by the smal1 quantities of Recuay ("Callején Unclassified") occurring. But his Figure 4 (the seriation chart) includes Castillo R/#/0 which occurred in what are apparently trace quantities at the Gallinazo Group and Castillo de Tomoval. Our survey found the Red/White/Orange style at Tomoval, V-316, V-322 San Ildefonso Sur, V-333 La Huaca, V-339 Co. Chirimoya, V-341 Gacho, V-345 Susanga, and V-346 La Toma (see Figure 5 and key). The highest frequencies then are at the valley neck, and up the Rio Huacapongo. These distributional details are interesting because an allied type (see Figure 22) is associated with the South Phase 2 style on the Carabanba Plateau, being especially common at Mollepuquia and Paredones.-18- The kaolin bowls of Phase 2 have been described in some detail in the earlier section. There are at least two styles of painted bowls, one coming from Cajamarca and one from the Callején de Huaylas. Detailed discussion of the bowls mist wait for completion of Steven Wegner's doctoral dissertation on Callején sites, and publication of the report of the Japanese Scientific Expedition to Nuclear America's excavation at Huacaloma outside Cajamarca. Cursory examination of sherds from the two areas indicated that there are significant similarities in the styles (thanks to Ryozo Matsumoto and Steven Wegner for their kind sharing of information). The Otuzco-Carabamba area lies PHASE 2 KAOLIN BOWLS Incised line Plain Painted Other MOCHE Siete Vueltas cruz Blanca Cruz Blanca cumbray Puente Larco cruz Blanca + 4 Cruz Blanca +4 — Co. Poroto cruz Blanca Co. Chile Dos de Mayo co. Chile Pedregal Dos de Mayo Co. Blanco del Leén Pedregal Bajo Co. Cucho Quebrada del Leén Panteén Co. Menocucho Santa Rosa Baja Huaca Menocucho Santa Rosa Baja Co. Blanco del Leén vIRU v-321 San Tlde- V-72 Castillode Sarraque V-74 Cast.Sarraque V-73 Cast .Sarraque ae V-339 Chimirmoya V-205 Corral Hill V-342 Que.Gacho V-992 Falda Calera v-332 Falda Calera V-333 La Huaca V-353 Co. Botador about midway between the two zones and it is not surprising to find the kaolin bowls represented both at sierra and middle valley sites. The incised bowls found on Phase 2 and 3 sites both in the sierra and in the middle valley are not represented in Cajamarca collections but have rough analogues in Callején collections. It is possible that they are of more southern origin, or are a local copy of a Callején style. A table (see above) shows sites at which kaolin bowls and found in the middle Moche and Viru Valleys. Frequencies in general are higher in Moche than in Viru. The single site with the highest number of kaolin bowls is cruz Blanca; at this site, bowls occur on the site proper (see Figure 4)19- but there are especially high concentrations on ridges adjacent to the main site, where there is very little preserved architecture. The next highest concentration, but in a burial context, was at Cerro Blanco del Leén. Kaolin bowls in Viru are concentrated in the middle and upper parts of the valley, with none occurring downvalley of Tonoval. Buff jars occurred on Co. Chirimoya, Vinsos Bajo, and Co. Pollino. A final cross-tie between Phase 2 sierra and mid-valley collections should be noted. Rim forms typical of the buff jars (Figure 13A and B) are very common in the middle valleys of both Moche and Viru, but done in the local sandy brown paste. Red painted lines usually occur, and their layout is identical to that on the imported buff jars. / Sierra sherds are far less frequent in the middle valleys during Moche TIT to V (equivalent to final Phase 2 and all of Phase 3). Some lined kaolin bowls occur, and some "plain" kaolin bowls, identical except for the lack of incision, in contexts that are probably Moche III-V. Control of the dating is not good however, and not much can be said except that there is apparently a major decrease in Frequency of coast-sierra interaction, as. evidenced by the ceramics. In Phase 4, sierra ceramics are again frequent in the middle valley. The rather restricted range of Phase 4 shapes are all found in the middle Noche Valley, espeically upriver from the La Cuesta-Hoche confluence. At a site like Co. Poroto, 90% of the collection has unmistable sierra affinities. Even across the river at Loma del Shingo, apparently a fortified Chimu town, there is substantial admixture of the sierra jars and bowls. The Phase 4 colander bowl with roughened exterior happens to be quite rare, occurring only at Co. Poroto and at Pante6n. This may reflect a basic difference in sierra and coastal diets, since graters never occur in the sierra but are fairly well represented in the middle valley, and colanders are almost exclusively sierra, where they ere quite frequent. Interestingly though, the colander bowl form does occur at many LIP middle valley sites: but without the perforations. Phase 4 flaring jars frequently have the single painted line at the lip interior, and the convex-flared jars often have conical or split-oval lugs. It is difficult to tell whether all of these sierra-looking vessels are imports or in situ manufactures, perhaps by Sierra craftsmen. Some vessels, for example those with gold mica incorporated into the paste, are probably made on the Carabanba Plateau and carried down. But the origin of the bulk of the sherds cannot be decided by visual inspec- tion. There are no sierra sherds in the m{ddle valley that can be attri- buted to long-distance trade. The situation in Viru is much the same. Sierra jars are common at sites well up the Rio Carabanba (Co. Lechuza, Mayasgo, Juyucul), where they co-occur with rather later imperial Chimu forms (cambers and angled plates). The sierra admixture and/or copying is more pronounced in the Carabanba drainage than in the Huacapongo. Co. Chirimoyo, the highest elevation surveyed site in the Huacapongo, had only a few Phase 4 sierra sherds, and the frequencies are even lower downvaliey. Obviously it would be desirable to have better control over the sources f the sherds discussed here, both long- and short-distance traded wares. It-20- is clear that some copying of sierra styles by mid-valley residents is occurring in both Phase 2 and Phase 4, and the Phase 2 serranos may be copying vessels made by their sierra neighbors to the north and south. To deal with this problem, it would be necessary to have far larger and better collections from all the possible source points as well as a good sample of destinations. An alternative though is X-ray florescent analysis of the constituent clay elements of sherds representing at least some of the styles and paste types discussed here. A Trent student, Michael Cawarno, has recently em- barked on such a study which will result in a Master's thesis. Very prelim- inary tests on 16 sherds, in which levels of several elements were measured, indicate that: a. Phase 4 sierra brown pastes show some clustering; b. three different kinds of kaolin paste shoved very weak clustering; c. the orange-grey Northern Phase 2 type does not cluster. The first step will be to identify the most easily traced paste types; this will probably require very narrow definition of styles, and location of clay sources. ‘Then comparison of samples of a given paste from as many sites as possible, both sierra and middle valley, may be able to confirm or refute assertions of exchange or imitation. CONCLUSIONS . The Forts Project is now into the analysis and synthesis stage. We will not attempt to present detailed conclusions here; the aim has been to make some of the data gathered in 1980 (and earlier) available to those with an interest in fortifications, in Central Andean prehistory, and/or in prehistoric interaction. The following list of observations and opinions will stand in Lieu of a summary. 1, In both coastal and sierra sections of the survey zone, there is a consistent tendency through time for single-site defenses to first become more sophisticated and then to give way to strategically located special-purpose fortifications. 2. In the sierra survey zone, higher elevations are favoured in the earlier part of the sequence, lower elvations in the final phases; this change may reflect increasing importance of food crops in the diet with a corresponding decrease in the proportional importance of herding as a food source, probably correlated with population increase. 3. The Inca presence is very difficult to discern. 4, During the Late Intermediate Period the most important sierra sites are located close stern edge Of the sierra zone, and are clearly engaging in exchange with the coast; it is not clear whether they are functioning more as producers or middlemen.-21- Each of the major Phase 4 sierra sites (Cuidista, Chamana, Carpaico) controls its own hinterland, and the three are probably in competition with one another. 6. The Noche-Viru middie valley area is best considered a buffer zone between coast and sierra, through which communication and exchange channelled, especially in Phases 2 and 4. 7. The mid-valley zone is clearly under Chima domination in the LIP, but in Salinar, Gallinazo, and MOche III-IV it is likely that mid-valley populations are autonomous; the Moche V situation is not clear. 8. Phase 2 exchange patterns are particularly interesting; mid-valley sites are probably controlling coastal access to sierra products.BIBLIOGRAPHY, Bawden, Garth 1977 ' Galindo and the Nature of the Middle Horizon in Northern Coastal Peru. Unpublished PhD Dissertation, Dept oF Anthropology, Harvard University. 1982 "Galindo: A Study in Cultural Transition during the Middle Horizon." In: Chan Chan: Andean Desert City, Michael Moseley and Kent Day, eds. University of New Hexico Press. Bennett, Wendell C. 1950 | The Gallinazo Group, Viru Valley, Peru.Yale Univeristy Publications in Anthropology No. 43, New Haven. Brennan, Curtiss T. 1980 | "Cerro Arena: Rise of the Andean Elite." Archaeology Vol. 33:3. Conrad, Geoffrey W. 1974 Burial Platforms and Related Structures on the North Coast of Peru: Sone Social and Political Implications. Unpublished PAD Dissertation, Dept of Anthropology, Harvard University. Donnan, Christopher 1976 Moche Art and Iconography. UCLA Latin American Studies Vol. 33, LA. Ford, James A 1949 "Cultural Dating of Prehistoric Sites in the Viru Valley, Peru.” Anthropological Papers of the American Museum of Natural History, Vol. 43 part 1, N.Y. Kosok, Paul 1965 ' Life, Land and Water in Ancient Peru. Long Island Univeristy Press. Mackenzie, Janet 1980 Coast to Highland Trade in Precolumbian Peru: Dendritic Economic Organization in the North Sierra. Unpublished MA Thesis, Dept oF Anthropology, Trent University. Moseley, Michael and Kent Day (eds) 1982 Chan Chan: Andean Desert City. University of New Mexico Press. Thatcher, John 1972 ‘Continuity and Change in the Ceramics of Huamachuco, North Highlands, Peru, Unpublished PhD Dissertation, Dept of Anthropology, University of Pennsylvania. Topic, John R. and Theresa Lange Topic 1978 "Proyecto Arqueolégico de Fortificaciones Prehistécias del Norte del PerG: Informe Preliminar sobre la Primera Temporada de Investigacién, dunio-Agosto 1977." Report to Instituto Nacional de Cultura, Lima. 1979 "Prehistoric Fortification Systems of Northern Peru: Preliminary Report on the Second Field Season, June-August 1978." Report to SSHRC.Topic, John R. and Theresa Lange Topic 1982. "Hiyamachuco Archaeological Project: Preliminary Report on the First Season, July-ugust 1981." Report to SSHRC. n.d, "Coast-Highland Relations in Northern Peru: Some Observations on Routes, Networks, and Scales of Interaction." ms. Topic, Theresa Lange 1977 | Excavations at Moche. Unpublished PhD Dissertation, Dept of Anthro- pology, Harvard University. Willey, Gordon R. 1953’ "preistoric Settlement Patterns in the Viru Valley, Peru." Smithsonian Institution Bureau of American Ethnology Bulletin no. 185, Washington DC.a oo ; p tah AS AE fea) - ‘ et ge Ley YAW eae ae » Ss a, eS De we i X if Set i ie peaFIGURE 2Fy 12 43 a KEY 10 FIGURE 2 Cerro Galindo (Chimu) Galindo (Moche V) Cerro Oreja Fort (Chimu) Frenteoreja (Gallinazo, Moche, Chimu) Quebrada Terrazas (Gallinazo Cerro de Terrazas(Gallinazo, Chimu Cerro Huérfano (Gallinazo, Chimu) Quebrada Riveros (Chimu) Cerro Blanco (Chimu) Puente Serrano (Salinar, Gallinazo, Moche V, Chimu) Santa Rosa Baja (EIP)* Cerro Petroglifo (Chim) La Granja (Gallinazo) Burro Muerto (Chimu: some EIP) 4sierra sherds on coastal siteSITE LOCATION--MAP 10 : x Ug = ES Ca Si, eRe RG oe 5a aAKEY TO FIGURE 3 Quebrada de Guitarras Cervo Blanco del Leén (Gallinazo, Moche, Moche V, Chimu)* Quebrada del Leén El Palmo (Moche ITI-IV) Santa Rosa-E1 Palno (Salinar, Galiinazo, Noche, chimu)# San Agustino (Moche) Cerro Cucho (late Salinar, Gallinazo, Chimu)* Muralla de Menocucho (Chimu lastest in wall) Cerro: Menocucho (Gallinazo, Moche, Chimu)* Panteén (Gallinazo, Moche, Chimu)* Pedregal (farly Chimu, Chimi)* Pedregal Baio (Noche, Moche V, Early Chima, Chimu)# Cumbray (Salinar, Gallinazo, Moche, Moche V, Chimu)* Riobar (Chimu)* Huaca Menocucho (Salinar, Gallinazo, Moche, Middle Horizon, Chimu)# Jesus Maria (Chimu) Jesus Marfa Village (Chimu) Dos de Mayo (Gallinazo, Chimu)* Cerro Chile (Gallinazo, Moche, Moche V, Chimu)# La Botija (Gallinazo, Moche, Moche V)* Cruz Blanca (Gallinazo, Moche V)# Cruz Blanca Divide #4 Cruz Blanca Divide #3 (Gallinazo) Cruz Blanca Divide #2 (Gallinazo) Cruz Blanca Divide #1 (Gallinazo) Huacamochal (Moche III-IV) Los Gentiles (Gallinazo, Moche, Chimu)* Pefia Larca (Chimu)* Puente Larco (Moche, Chimu)* Loma del Shingo (Chimu)* Siete Vueltas (Moche V, Chimu)# Cerro Poroto--various sections (Chimu)* Co. Cauro--sierra site, Phase 2 El Alto--sierra site, Phase 4 Major sierra sites surveyed in previous years 35 36. 37 38 39 40 44 42 43 44 Shamana Phase 4 Co. Quinga Phase 2 Cuidista Phase 4 Sulcha Phase 2 Paredones Phase 2 Mollepuguia Phase 2 Pefia Blanca Phase 2,4 Pefla Negra Phase 4 Carpaico Phase & Rogoday IncaFIGURE 4 CERRO CRUZ BLANCA Gre DPB rau akePIGURE 5 SETE LOCATION--MAP 11KEY TO PIGURE 5 Underlined sites are new sites, not included in Willey (1953). New sites are dated, as follows: Pu=Puerto Moorin, WsHuancaeo, MH=iddle Horizon, Ch=Chimu. on site surface 1 Castiuio del Inca 2 Co. las Pifias* 3 Co. Sausalito 4 Co. Liquen PH 8 Huancaco 6 Chuquil Bajo 7 Chuquil 8 Bitin 9 Palacio de Bitint 10 Bitf2 Bajo 41 Santa Clare 42. Viru Viejo 13 Napo 14 Tomoval 15 San Tidefonso Sur PM, G 16 San Francisco C,H 7) VCO Rees 18 v-52 19 San Ildefonso 20° Sarraque* 21 Castillo de Sarraque* 22 Castillo de Sarraque SE 23 V-316 24 Co. Pongo% 25 Co. Guanabana* PM,G,MH,Ch 26 Go. Pacffico ch 27 V=318 G,Ch® 28 San Nicolas PM,G,H,Ch* 29 Vinchos Bajo #,CH,PM,G 30 Paso Quebrada (walis)’H,Ch 31 Vinsos Bajo P¥,G,H,Ch® 32. Co, Nareos PMG 33 Susanga G,Ch 3% Chao Pass Walls G,Ch 35 Terrazas Le6n Ch 36 Gacho G,ch® 37 Co. Chirimoyo PM,G,Cht 38 La Tona PH,G,ch 39 Nifio Alto 6,Ch 40 Huacapongo G,ch allinazo, * indicates sierra sherds found M1 42 43 44 4s. 46 47 48 49 50 St 52 53 Su, 55 56 87 5B 59 60 61 62 63 64 68 66 (67 68 69 70 mn 72 73 7 8 6 7 Nifio Bajo 6,Ch Frente Vinsos PM,G,H,Ch Québrada_e} Nifio PH,G,ch ton Willey G,Ch Co, Pollino G,ch# Co. Botador PM,G,Ch* Quebrada Gudarra’ PM,G Corral Hill Quebrada Tomoval PM,G,H,MH,Ch V=1a5 Vista Confluencia PM,G,Ch* San Juan SE San Juan NW Que. de 10s MGsicos MH,Ch® Quenetite ,Ch V-1908 Quebrada Seca PM,G,H,Ch Genisal PM,Ch Que. Las Salinas ch Puguio Grande PM,G,Ch Los Trés Condores’ G.Ch* La Huaca PM,G,F,Ch* Go. La Calera Ch Falda de Calera 6,Ch* Co. Lechuza PN,G,H,HM,Ch,Inca® Mayasgo P¥,G,ch* To. de la Cruz (Chao) LIP) v-25 v-29 v-30 v-31 v-147 Huaca 1a Gallina Huaca e1 Gallo Co. Juyacul 6,Ch* Quebrada Cerro’Gacho PM,G,H,Ch* Cerro Coronado (Chao) Ch)FIGURE 6 SITE LOCATION--MAP 2 Key: 1 Co. Pan de Azucarae ce % BY C Sa 1 s ee ae : hate iD Co. Pongo (Vira)SITE LOCATION--MAP 1EIGURE 11 Phase 2, Northern Style, common rimsFIGURE 12: Phase 2, Northern Style-—typical jar form/decorationFIGURE 13 Phase 2, Southern Style, common rimsFIGURE 14 Fine Bowl Rims (mostly Phase 2) B is half-size, others 1:1.FIGURE 15 A Colour Key B Mollepuquia 224681. 10 on or © — MoliepuquiaFIGURE 17 Probable Cajamarca Pre-Cursive Bowls A Crug Blanca: Brown and white/grey B Cruz Blanca: Brown/orangeFIGURE 18 Kaolin Bowls with Incised Line A MoLlepuquia B — Mollepuquia C Co, Pedregal\r -FIGURE 20 Phase 4 Jars A Carpaireo B CarpaicoPRONEK MASSAyy ks SAY Paredones LLYLPYFIGURE 22 Red/white/Orange Sherds--Sierra Paredones B Paredones c Penia BlancaCERRO LOMA DEL SHINGO FIGURE 23