1If there are lessons to be learned from ancient agricultural practices, we first must reconstruct these practices so that we may compare them to contemporary situations. Archaeological research offers the only hope of reconstructing economic and social practices of people who left no written records or whom history has ignored, including the vast majority of all farmers who ever lived in Yemen. Since they cannot directly observe non-living people, archaeologists must rely on analysis and interpretation of material remains to infer the practices of ancient farmers.

2The RASA Archaeological Project, dedicated to exploring the Roots of Agriculture in Southern Arabia, seeks to document the introduction, implementation, and change in domesticates and agricultural technologies in the southern Jol. This paper discusses some early results of RASA archaeological research and explores theoretical problems in applying this archaeological knowledge to contemporary situations.

3Archaeological analysis relies on both technological and typological studies of material remains. Typological study, necessary for classification and characterization of ancient cultures, examines the attributes (form, metrics, raw materials, construction, style) of the end-product of material production: examples are the type of stone tool, or type of plough, or type of irrigation network, or type of crops produced. Technological studies (sometimes characterized as chaînes opératoires) define the technological steps necessary to achieve production. These technological steps form the basis for comparison of production strategies between cultures, places, or periods. If one anticipates applying archaeological knowledge of ancient agriculture to modern contexts, it is critical that one understands how that archaeological knowledge is produced—whether through typological or technological analysis. This paper will argue that use of typological results in modern agrarian development inherently carries considerable risk of failure, while use of technological results from archaeological research offer theoretically and practically sound application.

Ancient agricultural practices in Hadramawt (results of RASA research)

4The RASA Project has focused on the southern uplands of Hadramawt (Jol), possibly among the first locations where food production could have been adopted in southern Arabia. Runoff water management and fertile sediments make it possible today to farm discrete pockets of the Jol, and the broad basins of Raidat al Marra and Kor Sayban support small permanent villages.

5Past climates were without question more clement. The early and mid-Holocene saw enhanced southwest Asian monsoons.1 Land-based records show heightened moisture with a gradual decline in precipitation across Arabia 5000-6000 years ago.2 It can be difficult to reconstruct the precise conditions ancient farmers faced at any given time and place. The RASA Archaeological Project has begun local high resolution paleoenvironmental studies that rely on local records such as plants preserved in rock hyrax middens, magnetic and sedimentation signals in wâdî silts, and sinkhole cores from the coastal plain.

6RASA research has focused on two areas—the Wâdî Sanâ and the upper drainages of Wâdî ^cIdm, both major north-flowing tributaries to Wâdî Hadramawt. In the Wâdî Sanâ, a reconnaissance survey carried out by the German Archaeological Institute in 1993 first located Neolithic sites. There are now many indications that Neolithic (8000-4500 BP) activity in this area had been relatively continuous and its traces relatively undisturbed. The confluence of Wâdî Sanâ and a minor tributary, the Wâdî Shumlya (Shumayliya), has remnants of extensive deposits of silt (here formed between 11,000 and 4500 BP) in which archaeological preservation of a Neolithic landscape is remarkable. Although we still have questions about the actual date in which agriculture was first practiced in Hadramawt, preliminary results point to a changing history of environmental conditions and agrarian adaptations.

7The Wâdî Sanâ was much wetter 6000 years ago. Dramatic erosion shaping the major wâdîs of Hadramawt dates to the Tertiary period, but abundant local evidence points to greatly enhanced rainfall in the early Holocene. The extensive alluvial silt beds containing Neolithic sites and surfaces are no longer forming. RASA has collected wood charcoals from sediment profiles that bracket sedimentation between 6000 and 4500 BP (uncalibrated: uncal. hereafter), and geomorphological studies show that both alluvial and aeolian processes account for their accumulation. Although not precisely dated, surface collections of unifacial blade-tool assemblages suggest that silt deposition may reach back to the beginning of the Holocene. Furthermore, OSL (optically stimulated luminescence) dates suggest at least 11,000 years ago for the earlier sedimentation3. Since RASA has not yet dated basal silt deposits, we may only presume that sedimentation roughly corresponds to the regional wet phase beginning as early as 13,000 years ago.4

8In addition to the alluvial processes that caused silt to settle in vast shallow, seasonal ponds of water at the confluence of tributaries, plant growth also points to more moist local conditions in the early Holocene. Intensive RASA survey has documented extensive surface areas (5600 m²) and abundant horizons in section that represent burned land surfaces. Magnetic and geomorphological examination indicate that they represent multiple massive brush fires or gallery forests that have burned in situ. RASA suspects these fires were set and maintained by humans as part of a land management strategy; the burned layers date to about 6000 BP (uncal.). Relative stratigraphic dating suggests they correlate across several kilometers of now-dissected terrain. In association with these traces of a moister landscape, RASA archaeological survey has located and excavated numerous hearths (6000 BP uncal.), open air occupations in front of rockshelters (around 7000 years ago), and what appear to be Neolithic pit houses. Faunal remains suggest that people kept sheep or goats, but one of the more astonishing finds is the burned bones of cattle in a hearth. Whether the beasts were hunted or domestic, their presence confirms that this landscape once offered moister conditions more suitable for denser human settlement and perhaps for farming.

9What evidence exists for ancient agriculture in the Jol? RASA can identify a number of different physical remains, possibly dating to different time periods and certainly related to different environmental, social, and economic circumstances.

Shruj (sharj-shirâj)

10Shruj agriculture captures slope wash from a relatively small catchment adjacent to agricultural terrain, concentrating runoff usually onto a single field. Because current rainfall is very low, shruj may function only sporadically and probably occupies a marginal role in economic strategies. Material evidence consists of low alignments of stone 1-2 cobbles high and seldom more than a single cobble wide, although more elaborate modern examples may consist of a well-constructed wall. A typical shruj runs across a slope and diverts slope wash toward a natural gully or rock channel feeding agricultural fields below. Many of the shruj observed by the RASA group, while undated, are clearly ancient: silt beds below them have become “marooned” by gully-forming erosion so that no water can be directed to these isolated blocks today. In some cases, agricultural check dams associated with shruj have been cut by erosion or have deteriorated from disuse.

Check Dams

11Check dams may be constructed in association with shruj, or they may be isolated in secondary gullies or stream beds. While construction may vary, the purpose of check dams is to slow spate water so that it drops a nutrient-rich sediment load and sinks into the underlying sediment. Because RASA has conducted extensive study of local sedimentation sequences, rates, and correlation, it has been possible to situate several archaeological structures—probably check dams—in secure temporal and environmental contexts.

12In the Wâdî Shumlya, two structures appear to have been most likely constructed as check dams. The first, largely destroyed by flooding more than 5000 years ago, was built on silts accumulated no later than 6000 years ago (OSL dates on sediment). The second structure, a check dam spanning a 200 m stretch across eroding silts, is partially buried in sediments dated between 8800 and 5800 years ago (OSL dates). Because erosional processes have replaced depositional ones, the uppermost layers of sediment have been lost and cannot be dated, but regional geomorphological studies suggest that sediment deposition effectively ended about 5500 years ago (4600 BP uncal.). These check dams therefore represent the oldest known water management structures in Hadramawt.

Rayy

13Called simply “irrigation” by local inhabitants today, this agricultural practice relies on canals that take off water upstream to feed raised silt beds at the margins of a wâdî through which spate water flows. The technique has been much described elsewhere.5 In the smaller tributaries of the Jol, rayy may simply consist of a canal that collects slopewash from one side of the wâdî and conducts it downstream (several hundred meters up to 1 km). Typical also of rayy are revetments, built to prevent erosion and undercutting of fields by spate waters in the wâdî channel. Therefore, archaeological evidence of ancient rayy agriculture includes both canals and revetments. While these may be difficult to date, such structures often exist on and within “marooned” silts, clear indication of their former rather than contemporary use. Rayy agriculture probably continued throughout a period of increasing aridification following the Mid-Holocene moist conditions.

Dams

14At the mouth of Wâdî Sanâ and within view of a substantial pre-Islamic settlement, the RASA project recorded an ancient, well-built, large irrigation dam. This structure most probably functioned similarly to the well-documented ancient irrigation systems at Ma’rib,6 Shabwa,7 and the Wâdî Dura’.8 Spate waters could be slowed and partially diverted to feeder canals distributing water on agricultural fields at the margins of the wâdî.

Aflaj

15Aflaj, locally called qanawat rayy are only evident along the coastal plain where sinkholes have exposed a relatively high water table that is tapped for permanent cultivation by underground canals. These canals have been excavated through limestone bedrock, and they carry water at gentle grade up to one kilometer underground before emerging at the surface at lower elevations. The antiquity of aflaj is unknown, but at Ghayl Bâ Wazîr where they are currently maintained and used, several off-takes at the sinkholes sit well above present water level. This evidence suggests either that some canals are fed by seasonally higher water table or that the system was used when water re-charging was higher in the past.

16The same basic technology, although requiring much less labor, appears at small oases and seeps along the coast. In such cases, a landowner may dig a shallow well, install a pipe in one side below water table, and dig a trough several hundred meters downhill to tap high water tables for date palm irrigation. Because fresh water floats on salt water, this simple technology can be used only a few hundred meters from the shoreline where fresh water tables are high.

A developmental typology of agricultural practice

17RASA’s first analysis has been to organize a working developmental typology from the evidence of ancient agricultural practices. Dating all these types of agricultural structures has proved to be difficult. Only in a few examples can one obtain a relative date. In the Wâdî Sanâ itself, agricultural structures recorded by RASA are conclusively not modern, although some types (shruj, check dams, rayy, qanawat rayy) continue to be used elsewhere in the southern Jol.

18There are clearly also environmental variables that affect peoples’ adoption of different agricultural practices9 so that all types cannot be organized into a purely chronological developmental sequence. Nevertheless, some of the social and economic attributes (observed where these practices are in modern use) can be organized on a gradient of increasing scale. This approach has provided us with a working hypothesis of chronological development that can be later tested by precise dating in the archaeological record.

19Shruj agriculture requires relatively low-labor investment, little negotiation with neighbors over water rights, and no commitment to sedentism. Contemporary shruj plots may or may not yield in any given year and are typically maintained by tribesmen either residing elsewhere or moving with flocks that occupy primary consideration in economic decision-making. Greater labor inputs are required for construction and maintenance of check dams and rayy, and as the scale of catchment increases, so does maintenance. Typically one finds at least temporary shelter constructed near functioning rayy canals.

20The RASA team has examined no dams in current operation, but the extensive literature on large-scale sayl irrigation manipulated by these structures suggests that they are the hallmark of complex social arrangements for distributing and allocating water and adjudicating land access. The labor requirements for constructing and maintaining large-scale irrigation networks considerably exceed those in simple rayy and shruj agriculture. Villages and former urban centers (none in Wâdî Sanâ) shelter farmers committed to the yields from agricultural fields.

21From these attributes of energy investment and sedentism associated with different types of agricultural constructions, the RASA Project has hypothesized a chronological development with shruj as an early adaptation followed by check dams, rayy, and finally dams and aflaj. The underlying assumptions have been that

• less complex systems developed earlier, and
• agricultural systems with high labor input and sedentism requirements, characteristic of Iron Age and early historical kingdoms in Hadramawt, must have had local precursors.

22These hypotheses have formed an appropriate basis for archaeological research in remote, upland areas such as Wâdî Sanâ, and our results as a consequence offer exciting new insights into early agriculture as we are increasingly able to date early water management structures. Nevertheless, as untested hypotheses, they carry some danger if adapted and uncritically employed in contemporary application of traditional farming knowledge to modern Yemeni agriculture. It is to this latter point that the rest of this paper will be devoted.

Ethnographic analogies and typological thinking: problems and solutions

23Recent critiques of typological thinking in archaeological practice have pointed to several persistent problems. One of these has been the arrangement of modern ethnographically-observed practices into a developmental chronology that effectively effaces historical contingency and culture histories.10 By collapsing many trajectories of cultural experience into a unified model of typological development, anthropologists risk emphasizing static systems over dynamic ones. Where stages of development (e.g., of agricultural practices, or of cultural complexity) have been constructed from diverse ethnographic cases, no common processes or historical factors link one case to another. Hence the dynamic processes that account for maintenance and transformation of cultural and agricultural practice have been dissociated from the model type. Boserup’s famous economic model of agricultural intensification suffered from this construction. One critique of her approach has been the fallacy of stringing disjointed cases from tropical and temperate agriculture into a single developmental sequence devoid of conjoining processes that would account for transformation from one stage to another.11

24Another problem with typologically constructed models of chronological process lies in an implicit disregard for cultural histories. Historical contingency, critical to the transmission and maintenance of particular cultural practices, plays no role in a typological chronology. Typological models make no allowances for “multi-pathing,” or multiple trajectories of possible development. An inherent progressivist thinking underpins typological models.12 The obvious pitfall lies in failure to recognize equally developed alternative cultural practices. In the case of agriculture, multiple developmental responses to different conditions and historical contingencies most assuredly characterized ancient Yemen, just as they characterize contemporary Yemen.13

25While typological models nevertheless play a valid and important role in archaeological research, archaeologists must rely on other methods to interpret and explain the dynamic changes evident in an archaeological record of ancient cultural and agricultural development. Technological analysis, which examines the tasks, labor allocations, and labor roles necessary to achieve production given the material means to do so, provides the appropriate interpretive framework for this challenge. Technological analysis allows archaeologists to recognize changes in socially mediated access to land (including Islamic and tribal land tenure) or other productive resources (e.g., water, draft animals), labor roles, and ideology when changes occur in the material conditions of agricultural production. Material changes—the introduction or disappearance of crops like tobacco and indigo, or the use of cattle or camels as draft animals—can be archaeologically detected with relative ease. With analysis of the requisite labor roles and scheduling implicit in any historical shift in material use, archaeologists can interpret the social processes of agricultural and cultural development,14 making the historical cases of Yemeni agricultural practices in the past of far greater relevance to modern planners.

Conclusion

26While there is no time to delve further into technological analysis, several general implications warrant emphasis. First, while typological models do play a valid role in developing archaeological research agendas, there exists an inherent circularity if the archaeological hypotheses are then uncritically applied in modern development schemes. If contemporary (ethnographically documented) agricultural types have provided a means to reconstruct ancient practice, then no new knowledge is gained by applying the ancient agricultural practice in modern contexts! Second, as in modern Yemen, ancient Yemen’s agricultural practices were firmly embedded in social contexts that are the goal of interpretation in archaeological studies. Only when explanation of ancient agricultural change has been generated—the result of technological analysis of archaeological remains—will there be new knowledge to bring to contemporary situations. Therefore, archaeological studies that focus on labor sequences, labor roles, and scheduling of production in different agricultural contexts will offer the greatest prospects for understanding both the fabric and the dynamics of ancient societies and for providing information useful to modern agricultural development. The lessons are new to no one in this conference, but they nevertheless bear repeating in an interdisciplinary forum. No one can expect to adapt techniques and practices from one culture to another without due attention to social context. And no one can expect that adopting ancient agricultural practices will not affect existing social relationships, such as gender roles, land ownership, and community participation in maintenance, that underpin agricultural production today.