Full chapter text
settlement
Floodplains and piedmonts
North Bihar and the Nepal Tarai combine Himalayan
piedmont, alluvial fans, active channels, wetlands, and older
floodplain surfaces. Settlement opportunity and archaeological
visibility are both conditioned by this geomorphology. A mound
may preserve repeated occupation while nearby low surfaces
are scoured or buried. The absence of exposed remains is
therefore not equivalent to absence of past population.
Mobility alongside village life
Fishing, herding, seasonal movement, woodland use, and
river transport continued alongside settled cultivation. Later
administrative sources often treated mobility as disorder
because it complicated taxation and surveillance. Prehistoric
reconstruction should avoid the same bias: mobility can be a
rational adaptation to variable water, pasture, and labour
demand.
The environmental caution
Neither flood nor drought is purely 'natural' in its historical
effects. Exposure is shaped by embankments, land tenure,
settlement policy, transport, and unequal access to relief. This
principle, documented clearly for the Kosi in the colonial and
modern periods, also guides interpretation of earlier settlement
without pretending that later institutions existed in prehistory.
A connected river system
The northern tributaries of the Ganga descend rapidly from
Himalayan catchments, cross the piedmont, and spread across
low-gradient plains. Their channels carry water and sediment
while shifting through time. The result is a mosaic of active
channels, levees, backswamps, abandoned courses, and older
surfaces. Communities could benefit from fertile deposits,
fisheries, transport, and moisture while remaining exposed to
erosion, avulsion, waterlogging, and crop loss. The historical
importance of the Kosi, Gandak, Bagmati, Kamla, and related
rivers lies not in a single unchanging course but in the
opportunities and hazards created by a mobile system.
Seasonality and the organization of work
Monsoon rainfall and river regimes shaped calendars of
sowing, harvesting, fishing, grazing, building, and movement.
Labour demand could vary sharply across the year, encouraging
households to combine cultivation with craft, transport, forest
use, or seasonal migration. Archaeology rarely preserves this
calendar directly, yet botanical remains, animal assemblages,
storage, house rebuilding, and site location can provide partial
clues. A settlement occupied repeatedly need not have had the
same population or function in every season and phase.
Piedmont, plain, and wetland
The Tarai and Gangetic plain should not be reduced to one
environmental zone. Piedmont tracts, forest margins, marshes,
river islands, natural levees, and older alluvial surfaces offered
different resources and constraints. Routes between these zones
could be as important as permanent residence within one of
them. The ecological variety also helps explain why a regional
archaeological record may contain both settled farming and
persistent fishing, hunting, herding, and gathering. Diversity of
subsistence is consistent with long-term adaptation rather than
evidence of a simple sequence from primitive mobility to
complete sedentism.
Visibility and the map of known sites
The distribution of recorded sites reflects modern
investigation as well as ancient settlement. Mounds near roads
and towns are easier to notice and revisit than low or buried
deposits. River erosion may expose one locality while sealing
another beneath sediment. Political boundaries, institutional
funding, rescue archaeology, and the history of survey affect
what is documented on each side of the India-Nepal frontier.
Empty areas on an archaeological map may therefore mark
limited research or poor surface visibility. Regional conclusions
should identify these survey biases rather than turning them
into claims of ancient absence.
Adaptation without environmental determinism
Rivers and soils constrained possibilities, but they did not
dictate a single social outcome. Communities made choices
through knowledge, labour organization, tenure, ritual,
exchange, and political authority. The same flood could renew
soil for one household, destroy land for another, and redirect
traffic toward a third settlement. Hill's history of the Kosi
demonstrates for later periods that state projects and property
relations redistribute environmental risk (Hill 1997). Used
carefully, that insight encourages questions about unequal
exposure in earlier times without projecting modern
embankments, revenue offices, or legal categories backward.
The foreland is an archive of moving sediment
The north Bihar plain and the Nepal Tarai form part of the
Himalayan foreland, a low-gradient depositional system in
which water and sediment move from mountain catchments
toward the Ganga. Modern geomorphological work
demonstrates that the rivers crossing this plain are not one
hydrological type. Sinha and Friend distinguished mountain-fed,
foothills-fed, plains-fed, and mixed-fed systems on the basis of
source areas, discharge, sediment characteristics, and channel
behaviour. Mountain-fed rivers such as the Kosi and Gandak
built large megafans, whereas smaller rivers in the intervening
tracts developed different muddy floodplain and interfan
environments. This classification is a modern process model, not
a direct map of prehistoric land use, but it is essential for
avoiding the assumption that every river corridor offered the
same settlement opportunities or hazards (Sinha and Friend
1994).
The surface visible today is also not a single-age landscape.
Channels migrate, levees accrete, flood basins collect fine
sediment, wetlands expand or contract, and older surfaces can
be buried beneath younger alluvium. A village mound may
remain topographically prominent while adjacent occupation
surfaces disappear under metres of later deposition. Conversely,
a newly exposed bank can reveal older deposits that were
invisible to surface survey. Environmental reconstruction
therefore requires two chronologies: the age and history of the
landform, and the age and duration of human activity upon it.
One cannot safely be substituted for the other.
Megafans and interfan plains are different archaeological
settings
The Kosi and Gandak are major Himalayan rivers with
extensive fan-shaped depositional bodies in the foreland.
Sedimentological studies show strong down-fan and subsurface
variability rather than a single uniform blanket of sand. Singh,
Parkash, and Gohain documented changes in Kosi channel
pattern and facies from proximal braided reaches toward
straighter and more meandering downstream reaches, while
later borehole, resistivity, and drill-core work showed that the
shallow subsurface architecture of the Kosi and Gandak
megafans varies markedly between proximal and medial sectors
(Singh, Parkash, and Gohain 1993; Sinha et al. 2014).
For archaeology, the implication is methodological rather
than deterministic. A coarse proximal deposit, a fine-grained
interfan flood basin, a natural levee, and a marsh margin differ
in drainage, burial potential, exposure, and the visibility of
artefacts at the surface. They may also have differed in past
agricultural or transport possibilities, but those social uses
require archaeological evidence. Geomorphology can identify
plausible environmental settings and preservation risks; it
cannot by itself identify a settlement, language, polity, or social
group.
Diagram 6. Himalayan foreland to Ganga: schematic landscape transect. The
diagram separates piedmont/Tarai, megafan or interfan plain, active channel
belt, and Ganga-margin settings while showing how the same alluvial system
can create resources, hazards, exposed sites, and deeply buried archaeological
surfaces. It is not to scale and does not reconstruct one prehistoric moment.
Original editorial diagram prepared for this book, 2026. © Gajendra Thakur,
2026. Analytical synthesis based on Sinha and Friend (1994), Singh, Parkash,
and Gohain (1993), and Sinha et al. (2014).
Source links: Sinha & Friend 1994 (Sedimentology DOI); Sinha et al. 2014
(Sedimentary Geology DOI)
The Kosi and the problem of the “westward sweep”
A familiar historical narrative described the Kosi as having
shifted progressively westward across its megafan over the last
two centuries. That formulation should no longer be repeated
without qualification. Chakraborty and colleagues re-examined
twenty-eight historical maps published between 1760 and 1960
together with satellite-derived geomorphology and near-surface
sediment exposures. They concluded that the mapped historical
channel positions do not support a simple, continuous westward
migration; instead, the pattern is better understood in terms of
oscillation, lobe switching, and nodal avulsion, with the 2008
eastward avulsion providing an important modern analogue for
large relocations (Chakraborty et al. 2010).
This correction matters for long-term regional history
because a dramatic modern reputation can otherwise become a
false master narrative for the entire past. Historical maps can
document channel positions for their own dates, subject to
cartographic uncertainty. They cannot establish the course of
the Kosi in the first millennium BCE or the second millennium
BCE without independent geomorphological dating. Conversely,
the recognition that the river system can reorganize abruptly
warns against treating any one modern course as an ancient
boundary. The appropriate historical statement is dated and
source-specific: a channel is mapped in a particular place at a
particular time, while older courses remain hypotheses until
independently tested.
The Baghmati shows why “river mobility” has several forms
The Baghmati provides a useful contrast because its middle
reaches in north Bihar occupy an interfan setting between the
Kosi and Gandak megafans. Jain and Sinha described a modern
anabranching system with frequent flooding, high sediment
load, gentle gradients, and repeated avulsions; their
reconstruction identified eight major avulsions across a broad
floodplain over roughly 230 years. The important conclusion is
not that the same events occurred in prehistory, but that
interfan rivers can be highly mobile for reasons and in forms
different from the large trunk-channel behaviour of a megafan
river (Jain and Sinha 2004).
A historical geography that labels every abandoned channel
simply “old Kosi” or treats every multi-channel reach as
braiding would therefore erase process differences. Braiding,
meandering, anabranching, cut-off formation, and avulsion
describe distinct but sometimes interacting behaviours.
Archaeological interpretation benefits from this vocabulary
because the mechanism affects where erosion occurs, where
overbank sediment accumulates, how wetlands form, and
where old channels survive as linear depressions. Yet channel-
form terminology remains geomorphic evidence; it does not
identify the people who lived beside a channel.
Subsurface architecture makes absence especially difficult to
prove
Electrical-resistivity soundings, borehole records, and drill
cores from the Kosi and Gandak megafans show substantial
vertical and lateral variation in gravel, sand, and finer alluvium
within the upper tens of metres. Sinha and colleagues
reconstructed proximal and medial differences and emphasized
temporal variation in hydrological and sediment flux. For
archaeology this is a reminder that the surface is a thin
sampling window across a much thicker Quaternary archive
(Sinha et al. 2014).
A surface survey can therefore be excellent evidence for what
is exposed while remaining weak evidence for what lies buried.
Deep alluvial burial is especially relevant to low mounds,
ephemeral structures, hearths, field surfaces, and small scatters
that do not create durable topographic relief. The reverse
problem also occurs: erosion and channel cutting can bring
redeposited artefacts to the surface far from their original
occupation context. A responsible regional map should
distinguish excavated in-situ contexts, surface scatters, chance
finds, and objects recovered from secondary river deposits.
Landform age and settlement age must be dated independently
An “older alluvial surface” is a geomorphic category, not a
direct archaeological period. People can settle an old surface
long after it formed, reoccupy it repeatedly, or abandon it while
adjacent younger deposits remain unused. Likewise, a young
overbank unit can bury an older settlement without making the
settlement young. Chronological arguments are strongest when
geomorphic dating, stratigraphy, radiocarbon or other scientific
dates, diagnostic artefacts, and site formation agree. They are
weakest when the apparent age of the landscape is simply
assigned to the artefacts found upon it.
This distinction also protects regional comparison from
circular reasoning. One should not date a floodplain surface
from an assumed cultural attribution and then use the surface
age to validate the same attribution. Independent dating and
explicit uncertainty are necessary if palaeochannels or terrace-
like surfaces are to contribute to the history of settlement.
Wetlands were neither wastelands nor permanent lakes
Backswamps, abandoned channels, marshes, seasonally
inundated depressions, and shallow groundwater zones are
recurring components of the north Bihar-Tarai alluvial
landscape. Their extent changes with monsoon intensity,
channel position, sedimentation, drainage, and later
engineering. Such environments can support fish, aquatic
plants, reeds, grazing, seasonal cultivation, and transport while
also imposing risks of waterlogging, disease, and unstable
access. These are ecological possibilities rather than evidence
that any named prehistoric community used every resource.
Wetlands also complicate archaeological preservation.
Waterlogged deposits can preserve organic material under
favourable conditions, yet repeated flooding may rework or
deeply bury occupation debris. Peat, pollen, phytoliths, diatoms,
molluscs, and sediment geochemistry can sometimes
reconstruct local environments, but these proxies require secure
sampling and specialist analysis. A literary memory of a marsh
or a modern wetland name cannot substitute for a dated
palaeoenvironmental sequence.
Seasonality links environment to labour without fixing one social
model
Monsoon seasonality creates recurrent but variable rhythms
of cultivation, grazing, fishing, transport, house repair, craft
production, and movement. A household might intensify one
activity during high water and another during the dry season;
another community might organize labour differently despite
occupying a similar landform. Archaeology can approach these
rhythms through crop seasonality, animal age profiles, storage,
house rebuilding, hearth use, and the timing of flood deposits,
but the surviving evidence is rarely complete.
Seasonal movement should not be treated automatically as
the opposite of settled life. People can maintain durable villages
while moving cattle, fishing camps, craft labour, forest
gathering, marriage parties, pilgrims, or trading groups through
wider territories. Mobility operates at different scales and
durations. The analytical question is therefore which persons,
goods, animals, or practices moved, along what routes, at what
season, and with what institutional constraints.
Microtopography can matter more than a regional average
elevation
On a very low-gradient plain, small differences in elevation
can affect drainage, flood duration, and access to a channel.
Natural levees, older alluvial remnants, dune-like sandy rises,
and anthropogenic mounds may provide locally raised ground.
Settlement preference for such features is plausible and often
observable in modern landscapes, but it must be tested
archaeologically. A mound can be the cumulative product of
occupation rather than the original cause of settlement, and
repeated rebuilding can create topography that did not exist
when the first inhabitants arrived.
For this reason, modern digital elevation models are valuable
for reconnaissance but should not be treated as archaeological
proof. Resolution, vegetation, buildings, embankments, roads,
and recent channel engineering can distort the microrelief
relevant to an ancient site. Field levelling, sediment sections,
coring, and excavation remain necessary when an argument
depends on decimetre- or metre-scale elevation differences.
The Tarai is an ecological gradient, not an ancient international
margin
The Tarai connects Himalayan foothills with the Gangetic
plain through changing belts of coarse piedmont sediment,
groundwater emergence, forest, grassland, wetlands, cultivated
land, and river corridors. Its modern division between Nepal
and India is critical for present law and administration but
cannot organize prehistoric ecology. Rivers, animals, plant
communities, sediment, and human movement crossed the line
that later became an international boundary.
The archaeological record is also institutionally divided by
that boundary. Survey traditions, museum systems, national
heritage laws, publication practices, road construction, and
rescue archaeology differ between India and Nepal. Apparent
contrasts in site density can therefore combine genuine
historical variation with differences in discovery and
documentation. Cross-border regional history should compare
methods and survey intensity before treating a blank zone on
one side of the frontier as an ancient demographic fact.
Climate and tectonics are mechanisms, not automatic
explanations
Monsoon variability, Himalayan sediment supply, local
subsidence or tilting, earthquakes, and autogenic river
processes can all influence channel behaviour. Modern studies
of the Baghmati and Kosi evaluate combinations of slope,
aggradation, sediment flux, and tectonic setting rather than
attributing every avulsion to one cause (Jain and Sinha 2004;
Chakraborty et al. 2010). This multi-causal approach should be
retained for earlier periods.
It is especially important not to use a historically documented
earthquake or a modern flood as a convenient explanation for
an undated archaeological break. A causal claim requires
chronological overlap and a plausible physical mechanism.
Where only a broad palaeoclimate trend or tectonic possibility is
known, the prose should say “may have influenced” rather than
convert correlation into event history.
Remote sensing expands the field of view but does not date a
feature
Satellite imagery and digital elevation models can reveal
abandoned channels, levees, fan lobes, drainage anomalies, and
large mounds across areas far larger than a trench or walking
survey. They are particularly valuable in a cultivated alluvial
plain where many features are subtle at ground level. The Kosi
reassessment by Chakraborty and colleagues demonstrates the
value of combining historical maps, satellite data, and field
sedimentology rather than relying on any one source class
(Chakraborty et al. 2010).
Remote sensing nevertheless identifies form more readily
than age. A sinuous depression may be a palaeochannel, a
modern drainage line, an engineered cut, or a composite feature
reused through time. A spectral anomaly can reflect soil
moisture, vegetation, crop choice, or archaeology. Ground
verification, stratigraphic observation, coring, artefact
association, and scientific dating are required before remote
features enter a chronological narrative.
From occupation to published site: the visibility filter
Between ancient occupation and the modern distribution
map lie several filters. Deposits may be eroded or buried; later
farmers may plough them; bricks may be robbed; modern
settlements may cover them; road or canal cuts may expose
them; surveyors may or may not visit; and only some
discoveries are excavated, dated, conserved, and published. This
chain means that “known sites” are a sample produced jointly
by past behaviour, natural taphonomy, and modern research
history.
Diagram 7. Why the map of known sites is not the map of past settlement. The
workflow shows how occupation evidence can be altered by burial and
reworking, later land use, discovery conditions, and publication, producing
preservation and survey bias in an active alluvial plain.
Original editorial diagram prepared for this book, 2026. © Gajendra Thakur,
2026. Analytical synthesis informed by the geomorphic and historical-mapping
methods of Chakraborty et al. (2010), Jain and Sinha (2004), and the
archaeological source-control rules used throughout this volume.
Source links: Chakraborty et al. 2010 (Quaternary International DOI); Jain &
Sinha 2004 (Geomorphology DOI)
A rigorous settlement study should therefore record the
denominator as well as the numerator: how much area was
surveyed, by what method, in what season, with what visibility,
and with what criteria for calling a locality a site. Repeated
survey can change the distribution dramatically. Apparent
“empty zones” should be labelled unsurveyed, poorly visible,
deeply alluviated, or genuinely low-density only when the
evidence allows that distinction.
Environmental risk is socially distributed
The physical probability of inundation is only one component
of historical risk. Access to raised ground, boats, stored grain,
kin networks, livestock shelter, labour, political protection, and
alternative land can determine whether the same flood becomes
a manageable seasonal event or a catastrophe. Hill
demonstrated this clearly for the colonial and modern Kosi,
where embankments, property relations, and state policy
redistributed exposure rather than simply eliminating it (Hill
1997).
That later evidence cannot be projected directly into
prehistory, but it supplies an important question: who had the
capacity to absorb environmental variability? Archaeology may
approach the issue through differences in house construction,
storage, settlement elevation, access to transport, diet, and
rebuilding. It should not assume a uniform community response
simply because households lived beside the same river.
Resilience without evolutionary ranking
A mixed economy of cultivation, fishing, herding, gathering,
craft, and mobility should not be described as an incomplete
stage on the way to fully settled agriculture. In a flood-prone
environment, diversification can reduce dependence on a single
field, season, or channel. The strategy can coexist with
permanent houses, social hierarchy, exchange, ritual
institutions, and political authority. Archaeological
interpretation should therefore ask what combination of
practices stabilized household reproduction under variable
conditions rather than rank communities by sedentism alone.
The same caution applies to settlement relocation. Moving a
village because a channel eroded its bank is not necessarily
political collapse; remaining in one place is not necessarily
stability if households repeatedly rebuild, import food, or lose
fields. Environmental history becomes historically useful when
it explains the range of choices available and the inequalities
among them, not when it turns river behaviour into a
deterministic cycle of rise and fall.
What Chapter 4 establishes
The strongest conclusion is not a single map of prehistoric
settlement but a method for reconstructing one. The Himalayan
foreland is a spatially variable, vertically layered, seasonally
dynamic alluvial archive. Kosi megafan processes, Gandak
megafan architecture, and Baghmati interfan dynamics show
that different rivers and landforms cannot be merged into one
environmental template. Modern geomorphology demonstrates
mechanisms and preservation problems; archaeology must
establish when and how past communities actually occupied
those settings.
Accordingly, later chapters use environmental evidence in
three disciplined ways. First, landform and channel
reconstructions are dated independently wherever possible.
Second, environmental mechanisms are separated from social
outcomes. Third, the distribution of known sites is treated as a
research sample rather than a complete census of ancient
population. These rules allow rivers to enter political, economic,
and cultural history as active material processes without
turning geography into destiny.
Chapter-specific bibliography
Hill, Christopher V. River of Sorrow: Environment and Social Control in
Riparian North India, 1770-1994. Association for Asian Studies, 1997.
Singh, Upinder. A History of Ancient and Early Medieval India: From the
Stone Age to the 12th Century. Pearson Education India, 2008.
Sinha, B. P. The Archaeology and Art of India. Sundeep Prakashan, 1994.
Sinha, Rajiv, and Peter F. Friend. 'River Systems and Their Sediment Flux,
Indo-Gangetic Plains, Northern Bihar, India.' Sedimentology 41, no. 4
(1994): 825-845. https://doi.org/10.1111/j.1365-3091.1994.tb01426.x.
Singh, Harbhajan, B. Parkash, and K. Gohain. 'Facies Analysis of the Kosi
Megafan Deposits.' Sedimentary Geology 85, nos. 1-4 (1993): 87-113.
https://doi.org/10.1016/0037-0738(93)90077-I.
Jain, Vikrant, and Rajiv Sinha. 'Fluvial Dynamics of an Anabranching River
System in Himalayan Foreland Basin, Baghmati River, North Bihar
Plains, India.' Geomorphology 60, nos. 1-2 (2004): 147-170.
https://doi.org/10.1016/j.geomorph.2003.07.008.
Chakraborty, Tapan, Rimpal Kar, Parthasarathi Ghosh, and Sounak Basu.
'Kosi Megafan: Historical Records, Geomorphology and the Recent
Avulsion of the Kosi River.' Quaternary International 227, no. 2 (2010):
143-160. https://doi.org/10.1016/j.quaint.2009.12.002.
Sinha, Rajiv, Jawed Ahmad, Kumar Gaurav, and Guillaume Morin. 'Shallow
Subsurface Stratigraphy and Alluvial Architecture of the Kosi and
Gandak Megafans in the Himalayan Foreland Basin, India.' Sedimentary
Geology 301 (2014): 133-149. https://doi.org/10.1016/j.sedgeo.2013.06.008.
VOLUME II.
ANCIENT MITHILA, VAJJI
AND ANGA.
Source-controlled regional history.