Kites in Saudi Arabia
It has become increasingly common in archaeology to utilise virtual globes for
regions where few if any aerial photographs are available. Saudi Arabia is one
such and it has proved especially useful for identifying and mapping the proliﬁc
structures commonly referred to as the ‘Works of the Old Men’, most prominently
kites. These are now generally accepted as hunting traps for migratory animals.
Although a few were known in Saudi Arabia, the increasing availability of high-
resolution ‘windows’on virtual globes has revealed them in ever-larger numbers.
Such windows can be exploited to deﬁne and map archaeological remains and
develop methodologies. One particular region with such potential is Harret Khay-
bar. Progressive additions of high-resolution windows for this harra have revealed
917 kites. Beyond mere counting, analysis allows the development of typologies
and identiﬁcation of locally distinctive forms—notably the ‘barbed’form; mapping
and the interpretation of patterns in relation to geology, soils, water sources and
vegetation; and associations with other ‘works’, with scope for creating at least rel-
ative chronologies. The present study provides data and preliminary analysis, and
guidance for such a holistic ground-based archaeological project should the oppor-
Keywords: kites, Kingdom of Saudi Arabia, remote sensing, satellite imagery,
Arabia, prehistoric, stone-built structures
, Matthew Dalton
University of Western Australia,
M204, Stirling HWY, Crawley
WA 6009, Australia
University of Cambridge,
Division of Archaeology,
Downing St, Cambridge, CB2
Kites have been known to western scholarship in various
parts of ‘Arabia’since their discovery and publication by
Royal Air Force (RAF) pilots in Transjordan during the
1920s. The articles by Maitland, Rees and Insall, all pilots
in the region, revealed a variety of stone-built site types
not previously known and apparently surviving in their
hundreds in the lava ﬁeld (harra) called Harret al-Shaam
of the Jordanian panhandle (Kennedy 2012a).
soon being reported by French pilots in the continuation
of that lava ﬁeld in southern Syria, and soon after in cen-
tral and northern Syria well beyond lava ﬁelds. Although
the border was then unmarked, it would have been clear
that strings of kites seen in Harret al-Shaam in Jordan con-
tinued south into the Arabian Peninsula. It might have
been expected that kites would also be found in some of
the similar harrat along the western side of the Peninsula
and—in the light of discoveries in Syria—more widely,
too, outside lava ﬁelds (Fig. 1).
The ﬁrst kite reported in the interior of Saudi Arabia
was recorded in 1951 (but not published until twenty-ﬁve
years later; Ryckmans 1976) (Fig. 2). Soon after, ground
surveys of the Comprehensive Archaeological Survey Pro-
gram (CASP) recorded kites in Wadi Sirhan (Parr, Harding
& Dayton 1968–69: 39), although the number was not
speciﬁed and none were mapped; another survey noted
two more in the Northern Province (Adams et al. 1977:
pls. 8 & 9). There matters rested for a generation, so that
Echallier and Braemer published their tally of all
kites in what they called the ‘Levant’, they could still
report only ‘a dozen’kites in all of Saudi Arabia (1995:
36). Plainly there were more, but the ﬁeldwork had not
been done. More to the point, the ﬂying and access to
aerial photographs that had been instrumental in the dis-
covery and the start of mapping in Jordan and Syria was
unavailable for Saudi Arabia.
The Arabic for a lava ﬁeld is harra; two or more is harrat; and
used in conjunction with a speciﬁc name it becomes harret —
e.g. Harret Khaybar.
Arab. arch. epig. 2015: 26: 177–195 (2015)
Printed in Singapore. All rights reserved
What we are here calling ‘Arabia’extends from the
mountains of south-eastern Turkey to the Arabian Sea,
about 3 million km
. Of that, over two thirds—some
—is the Kingdom of Saudi Arabia. Those
engaged in the CASP during the 1970s (see above) noted
the limitations imposed by having few if any aerial photo-
graphs (Kennedy 2011: 3197; Kennedy & Bishop 2011:
1285). Indeed, the only practical and cost-effective way to
explore such an extensive and geologically and environ-
mentally challenging landscape is through aerial survey.
As we know from interpretation of old aerial photographs
of, and from eighteen years of aerial reconnaissance in,
Jordan kites that survive on the surface are exactly the
kind of site that is readily visible from above.
A few of the aerial photographs taken in the course of
the Arab Revolt in 1917–18 survive, but are very limited
in scope and show nothing archaeological. A few of use
for current purposes were published many years later
along with at least one sheet of the 1:250 000 geological
maps (Geoscience Map GM-131) of Saudi Arabia in 1991
(see below). Now we can add 476 vertical aerial survey
photographs among the c.1.5 million in Rhodes House in
Oxford, taken in 1974–76 for the Directorate of Overseas
Surveys (see below).
This last is a useful addition but still
woefully small by any standards and especially for such a
vast landscape. By comparison, just across the border in
what is now Yemen, the British aerial survey was exten-
sive and the same aerial photo archive in Rhodes House
includes 45,361 images of that country. This may be an
undercount and there may be more aerial photos of Saudi
Arabia (and Yemen) in the other huge collection of c.1.5
million images being catalogued in Edinburgh as part of
the National Collection of Aerial Photographs
It is known from reports that large numbers of aerial
photographs have been taken for mapping purposes in
Saudi Arabia and more still in connection with geological
prospection. Almost none were available to the archaeolo-
gists of the CASP and they are still rarely available. Such
archives must number in the hundreds of thousands.
A map showing the location of the major lava ﬁelds in the Arabian
Peninsula. Harret Khaybar is in the centre (drawing: Matthew Dalton).
A sketch plan of a kite recorded in western Saudi Arabia in 1951 (from
Ryckmans 1976: 162).
Not 112 as on their spreadsheet.
To date (2 December 2014) only about 3500 (of mainly Kenya
and Tanzania) have been digitised and displayed online.
D. KENNEDY ET AL.
The major development—to some extent allowing
scholars to bypass aerial photographs—has been the publi-
cation online of virtual globes. Google Earth and Bing
Maps offer mosaicked satellite imagery covering the entire
planet, providing an immense photomap. The quality var-
ies but for several years the number of high-resolution
‘windows’being uploaded to Google Earth and Bing
Maps for Saudi Arabia has been increasing rapidly. The
results have been dramatic for the archaeology of Saudi
Arabia (Kennedy & Bishop 2011) and may be illustrated
very effectively by focusing on just this single site-type—
B. Kites in northern Saudi Arabia
A generation ago, only about a dozen kites had been
reported in Saudi Arabia and none had been speciﬁcally
located, much less mapped and recorded. By 2011, inter-
pretation of Google Earth had revealed sixty-seven, two
years later Kempe and al-Malabeh (2013) published a total
of forty-ﬁve, but the Aerial Photographic Archive for
Archaeology in the Middle East (APAAME) team had
already located 107. In May 2014 as a result of further
high-resolution windows being uploaded, we had a count
of 122, now increased (December 2014) to 136 (Fig. 3).
As Figure 3 shows, the kites in the north of Saudi Ara-
bia are largely the continuation of the chains of kites
recorded in the same lava ﬁeld in Jordan over a number of
years. In design they are very similar to those across the
frontier in Jordan, as one would expect. Furthermore, as
with many in the neighbouring part of Jordan, they are in
areas where light sand has largely covered them. This
A distribution map of kites in Harret al-Shaam as of 01/12/2014 (draw-
ing: Rebecca Banks and Matthew Dalton).
A kite in Jordan near the Saudi border, of a type visible on virtual globes
for northern Saudi Arabia but difﬁcult to locate at ground level;
Nukheila Kites 4 and 5 (APAAME_20081102_DLK-0211).
Comparison of a kite in northern Saudi Arabia: 3731-12-Jabal Nu’ayj
Kite 2 on Google Earth (top) and Bing Maps (below).
KITES IN SAUDI ARABIA
makes their identiﬁcation on the ground almost impossi-
ble, and thus very difﬁcult, through satellite imagery.
Examples from the Jordanian side, where we have low-
level aerial photographs as well as satellite imagery, make
this point and underscore the vital role of the aerial and
space view (Fig. 4).
Also of signiﬁcance here is the opportunity and need to
utilise Google Earth alongside other virtual globes. There
are parts of this area, which are of low resolution in
Google Earth but high resolution in Bing Maps (see e.g.
Fig. 5). Moreover, sites may appear clearer on imagery of
different dates or types. A remote-sensing survey of the
region therefore can and should utilise more than one
source of imagery. The opportunity to do this freely is pos-
sible due to competing online virtual globes and maps
(Google Earth and Maps, Bing Maps, Nokia’s HERE).
From this varied source material locational data can be
transferred to Google Earth or into Geographical Informa-
tion Systems (GIS) for analysis.
C. Kites in Harret Khaybar
The geologists who prepared the ‘Geologic Map of ...
the Harrats Khaybar, Ithnayn, and Kura ...’(1:250 000
Geoscience Map GM-131) used aerial photographs and
would have known of the kites visible in this imagery.
These photographs have not generally been available to
archaeologists and it was only when the map (and the
accompanying notes) was published (Roobol & Camp
1991) that speciﬁc reference was made to kites and an
The ﬁrst high-resolution window for Harret Khaybar on Google Earth (dated 19 February 2007); (left) the ‘window’in the context of Harret Khaybar
as a whole; (right) detail of the ‘window’showing the kite distribution (drawing: Rebecca Banks).
A map of Harret Khaybar showing the location of high-resolution win-
dows available on Google Earth as of July 2013 for the oases of Khay-
bar and Al-Ha‘it (drawing: Rebecca Banks). Cf. Figure 9 below for
signiﬁcant further extensions.
D. KENNEDY ET AL.
aerial photograph included. As the map had a restricted
circulation the discovery was not widely known.
The breakthrough came with the appearance online of
the ﬁrst high-resolution imagery of parts of Saudi Arabia
on Google Earth. In 2008 Dr Abdullah al-Sa’eed, a medi-
cal doctor in Riyadh with an interest in archaeology, drew
attention to an area of such high-resolution imagery (dated
February 2007; Fig. 6) near the town of Khaybar in west-
central Saudi Arabia and to the numerous stone structures
visible in it (Kennedy & al-Sa’eed 2009). Khaybar lies on
the western side of Harret Khaybar. Most striking were the
A systematic interpretation in 2009 of that window,
which covered c.2000 km
, recorded an immensely rich
harvest of 239 kites, many of them of a form not encoun-
tered in the heartlands of kite discovery in Harret al-Shaam
in the far north (see Fig. 3). The boundaries of that ﬁrst
Khaybar window were the arbitrary ones determined by
the availability of high-resolution imagery that Google
had recently uploaded. Even at that time (2009) the num-
ber of kites recorded represented a greater density (one per
in the c.1200 km
section covered by the lava ﬁeld)
than found even in the thickly marked Safawi area of Jor-
dan’s Harret al-Shaam (with one kite per 4 km
It was likely that kites would be found in the areas to
the west and south of that initial window when high-reso-
lution imagery of those parts of the lava ﬁeld became
available. One such new window did indeed reveal more
kites –taking the total to 297 by September 2011. By
2012 the total had been increased again from the upload of
Mapping kites and the growth in numbers; (left) high-resolution windows of the western part of Harret Khaybar around Khaybar town superimposed
on the 1:50 000 map grid of the area; (right) the same map grid showing the numbers of kites identiﬁed in each ‘map’area—the number in each case
represents the number of kites in September 2011 and the number (in brackets) in July 2013 (drawing: Rebecca Banks).
KITES IN SAUDI ARABIA
further high-resolution windows by Google, including the
discovery of ﬁfty-ﬁve around Al-Ha’it on the eastern side
of the harra (Kennedy 2012b: 147) (Fig. 7). A total over-
all in this harra is currently at least 917 (see Fig. 10, left).
The better quality of imagery now available for the initial
study window has meant that the kite total for that area
alone has risen to 311.
In undertaking a further and much more extensive inter-
pretation of the western part of the Khaybar lava ﬁeld, the
area around Khaybar town was gridded (as throughout
‘Arabia’as a whole in our wider research) on the basis of
the Saudi General Commission for Survey 1:50 000 map
sheets. This particular study extended over a grid of
twelve maps (Fig. 8) (NB: one of those maps is ofﬁcially
‘3925-42 –Khaybar Sheet’and not to be confused with
‘Harret Khaybar’as a whole).
Figure 8 (right) tells the story: the numbers in each
square represent the number of kites found in the area of
the Khaybar window when interpreted in September 2011.
Some squares contain zero kites because the area was not
covered at that time by a high-resolution window. The sec-
ond number (in brackets) is the number of kites recorded
by July 2013 after the interpretation of new high-resolu-
tion extensions. For example, sheet ‘3925-43 –Al Wadi’
had twenty-six kites in the original research in 2011 but
thirty-two in 2013. Much more noticeable, however, is that
the total number of kites in this group of squares rose from
297 to 575. Most striking of all is that the Khaybar sheet
—‘3925-42 –Khaybar’—rose from 75 to 291 kites. To
put this in context, previously by far the most densely pop-
ulated map sheet for any part of ‘Arabia’was that for
Safawi in Jordan with exactly 200 kites. At a stroke, this
Khaybar sheet now has almost 50% more kites than its
Much more high-resolution imagery has since been
uploaded, with several windows overlapping one another
in the vicinity of Khaybar and Al-Ha’it towns and much
of the entire harra now covered (Fig. 9). As of the 1
December 2014, our analysis of the imagery for Harret
Khaybar as a whole now counts 917 kites—772 in the area
nearest Khaybar town and a further 145 on the eastern side
and especially around the Al-Ha’it oasis (Table 1; Fig. 10,
As in similar terrain elsewhere in ‘Arabia’as a whole,
there were no visible kites in the mountainous northern
half of the original study window; more surprisingly there
were no kites (or any other structures) in the eastern part
of the window, towards the central area of the lava ﬁeld
(Fig. 6). The distribution of kites in Harret Khaybar is
very pronounced, with major clusters in the vicinity of the
oases of Khaybar and Al-Ha’it and a few outliers around
the south-eastern and south-western fringes of the lava,
but none in between (Fig. 10, left).
Although we do not have high-resolution imagery for
the entire area of Harret Khaybar, it is now very extensive
Table 1. Numbers of Kites in each part of Saudi Arabia as of 1 Decem-
ber 2014 (compiled by Rebecca Banks).
Saudi Arabia Kites
Harret al-Shaam 136
Harret Khaybar 917
Harret Rahat 19
Harret Kura 2
Khaybar East 145
Khaybar West 772
A pattern of high-resolution windows available on Google Earth as of
December 2014. Areas indicated in red are Al Huwayyit (right) and
Jabal Muzahim (left) (drawing: Rebecca Banks).
Recently, Kempe and al-Malabeh have explored the same harra
but found considerably fewer kites and made some puzzling
observations about them (2013: 138–139; see below).
D. KENNEDY ET AL.
(Fig. 9) and provides near certainty that there are indeed
no kites in the great central expanse. Nor is this lack of
kites due to their obliteration by recent Holocene volcanic
activity; the areas covered by more recent lava ﬂows are
relatively restricted (see below).
Although there may well be some further kites revealed
in the area south-east of Khaybar and south of Al-Ha’it as
new windows are uploaded, it seems unlikely that any
kites exist in the entire central part of the harra between
the two areas currently plotted around Khaybar and
Maps showing (left) the distribution of kites in Harret Khaybar; (right) the distribution of pendants (drawing: Rebecca Banks).
A chain of kites in the Jabal Muzahim area of Harret Khaybar (drawing: Rebecca Banks).
KITES IN SAUDI ARABIA
Al-Ha’it. Analysis of this central area shows that other
types of stone structure (see Appendix below) are abun-
dant, not least ‘pendants’which are found in immense
numbers everywhere (Fig. 10, right).
The pattern as currently presented in both the eastern
and western zones of Harret Khaybar is common through-
out ‘Arabia’. Kites are often found close to seasonal water-
courses (wadis), and are frequently arranged in
interconnected chains (contra Kempe & al-Malabeh 2013:
138–139). These chains are far shorter than many of those
in Harret al-Shaam (see Fig. 3 for an impression of these
networks’immense length in Jordan), but nonetheless
often have tails linked together (Fig. 11). These shorter
chains are probably a factor of different geographies and
seasonal water ﬂow patterns. In Harret al-Shaam, several
major wadis run for very long distances. In Harret Khay-
bar, wadis radiate outwards from the high ground and the
volcanic cones and vents, which occupy its centre (see
wadis indicated on distribution maps: Figs. 6, 7, 10 & 27).
It may be no coincidence that kites are concentrated at the
points where various wadis come closer together or merge
at Khaybar and Al-Ha’it oases, perhaps reﬂecting the
attractiveness of concentrations of water to the fauna of
the region. We may note, too, that the area where more
kites are beginning to appear on some very recent high-
resolution imagery is where another group of wadis con-
verges in the vicinity of Al Huwayyit (Figs. 9 & 10, left).
Of the kites analysed in the original study window
around Khaybar town, approximately 70% (n = 226) are
directly associated with a wadi. A similar pattern is
observed in the main Al-Ha’it windows, where 65%
(n = 47) of kites are built next to wadis. These characteris-
tics seem to stem primarily from the use of and adaptation
to local landforms, especially ridges (natural barriers or
tail walls) and passes (natural ‘funnels’) where a wadi has
cut through the landscape. The position of kites’funnels
or tail walls in low-lying areas such as wadis and/or mud
pans is signiﬁcant: when dry, these features would have
provided an easier surface to traverse, and therefore a
more likely thoroughfare for migratory game. Further-
more, these wadis are far more richly vegetated than the
surrounding desert, and thus would have very likely
attracted grazing herbivorous animals. There is also a
strong tendency for kite heads to be positioned at higher
elevations in relation to their tails, especially above ridges
or behind low rises. This positioning would have advanta-
geously rendered any hunters waiting there invisible to
animals already inside the kite.
RELATIONSHIPS AND ASSOCIATIONS
The Khaybar area includes a type of site not found in Har-
ret al-Shaam or further north; indeed, they seem to be
unique to west-central Saudi Arabia, though no longer just
to Harret Khaybar (see Kennedy 2011: 3193). These are
A diagram illustrating the direction of the opening of the kite ‘funnels’in
the Harret Khaybar original study window (created by Rebecca Banks).
A‘barbed’kite (Samhah Kite 15) (drawing: Rebecca Banks).
D. KENNEDY ET AL.
the structures called ‘gates’, which consist of two short
thick walls or heaps of stones, between which one or more
connecting walls stretch. From above, these features
resemble an old-fashioned barred gate laid ﬂat. In the ori-
ginal study window at Khaybar, 106 ‘gates’have been
recorded, but they tend to be concentrated in groups and
largely apart from kites (see Fig. 27).
The other category of structure found in abundance in
this harra is almost certainly funerary—cairns, pendants
and the very remarkable ‘keyhole’structures (which seem
to be an elaborated form of pendant) (see Kennedy 2011
and Appendix below). The latter two types are often
aligned on tracks radiating outwards from these oases. As
we shall see, kites do appear in the same area but rarely
intersect (see Figs. 14, 19 & 20–22), and these funerary
structures extend much further into the harra than kites do
The direction in which the funnel of the kite opens may be
of signiﬁcance in some cases. It has been interpreted else-
where as an indicator of the direction from which migra-
tory herds would arrive (Betts & Yagodin 2000: 40). In
Harret al-Shaam, kite funnels overwhelmingly open to the
east and east-south-east; in the western zone of Harret
Khaybar it is to the west and west-south-west (Fig. 12).
The directional trend of the Harret Khaybar group from
the original study window is strongest in those kites ‘net-
worked’in chains across the landscape. The only ‘charac-
teristic’that sees a departure from this preference appears
to be for those kites built alongside or associated with a
wadi or mud pans. Here the numbers show a weaker pref-
erence for a westerly funnel direction, indicating that the
placement of the kite in a particular landscape plays a
determining factor in its orientation. This correlates with
what is seen with the kites around Al-Ha’it. In this area,
kite chains also commonly share tail orientations, and here
this is also primarily a factor of their geographical place-
ment. Several chains of kites have been built on both sides
of the large wadi that runs into and then out of the oasis
itself. These kites are overwhelmingly positioned so that
their heads are hidden over ridges, while tails extend down
into wadi beds (see above). The mixed orientation of their
tails implies that they were intended to catch animals both
entering and exiting the oasis.
When they are not part of a chain, kites are more likely
to have distinctive characteristics: for example, they are
short; have fewer tails; have an irregularly formed head; or
the direction of their funnel is less likely to conform to the
strong orientation trends of the group.
The form of the kites in Harret Khaybar has some overlap
with those found in Harret al-Shaam, but most are distinc-
tive to this region. The most striking is the ‘barbed’head
type already identiﬁed in 2009 (Kennedy & Al-Sa’eed
2009) (Fig. 13) but there are other types (contra Kempe &
al-Malabeh 2013: 139 who seem to regard them all to be
of this form).
The term ‘barbed’seemed especially apposite for the
ﬁrst kites identiﬁed in 2009, which had the form of a
barbed arrow or spearhead. Of the kites in the original
study window, 85% can be conceived of as having ‘barbs’
on the head similar to that of a typical arrow ‘type’: two
blinds extending back either side of where the tails meet
the head, often ending in one or more circular hides. A
universal grouping of all these kites as ‘barbed’or ‘arrow’
is not meaningful, however, when the variation in form
aside from this one feature is considered (see provisional
typology below; Fig. 15). Nevertheless, the term remains
valid. The feature extends from the head to the tails; blinds
can be located along the length of the guide-walls, where
they often appear at intervals as small triangular barbs. In
some cases the ‘barbs’are larger and almost act as second-
ary ‘heads’. It is occasionally difﬁcult to discern where a
kite’s head ﬁnishes and its tails begins because of the
extent of the barbs along its guide-walls, as on Khaybar
Kite 12 (Fig. 14). Although kites with tail barbs are
characteristic of this region, the type is by no means the
A kite with tail barbs (Khaybar Kite 12); note also the proximity of the
pendant avenue (drawing: Rebecca Banks).
KITES IN SAUDI ARABIA
Provisional typology of kite types in Harret Khaybar (drawing: Rebecca Banks).
The average overall length of kites by type in the original Khaybar study window (created by Rebecca Banks).
D. KENNEDY ET AL.
majority—only c.25% of the kites analysed from the origi-
nal study window have these barbs.
A provisional typology graphically sets out the kite
types encountered in the Khaybar study window (Fig. 15).
The Y-kite—unique to Saudi Arabia and proliﬁc in this
study window—has as one of its main features a profusion
of hides located at the tip of its head and at the end of its
‘barbs’. Of note in this study group is the inclusion of
hides on kite tails as well; this is extremely uncommon in
the kites of other regions. Again, the Y-kite is the out-
standing example of this feature.
The size of kites in the region varies considerably (see
Kempe & al-Malabeh 2013: 139). The heads of the kites
can be large, c.5000 m
+ on average in the original
study window; the average length of kite tails is quite low
at c.350 m (Fig. 16), but the range extends from no tails
at all to a maximum of 1650 m in length. It is also impor-
tant to note that kite tails in Khaybar do not meander
across the landscape as they commonly do in Harret
al-Shaam, the latter beginning as a clearly deﬁned kite
wall but often developing into a ‘meandering wall’or a
tangle of walls at its full extent (see Kennedy 2011: 3190).
Kites and associated structures in Harret al-Shaam (Safawi Kites 109–
110, 120) (drawing: Rebecca Banks).
A common pendant and wheel relationship in the Jordanian part of Har-
ret al-Shaam (Safawi Wheel 11 overlain by Safawi Pendant 4) (APA-
Al-Silsilah Kite 50 overlain by Pendant 6 on Google Earth (drawing:
KITES IN SAUDI ARABIA
The trend is for more structurally complex kites (barbed,
rectilinear, star and Y-kite) and those kites in chains and/or
with tail barbs to be longer.
The evidence for the dating of kites in general has been
summarised several times recently; overall, there is very
little secure evidence to work with (Kennedy 2012b: 145
with references; cf. Betts 2014; Zeder & Bar-Oz 2014).
RELATIONSHIPS AND ASSOCIATIONS
In Harret al-Shaam, there are numerous instances of kites
close to or physically intersecting with other ancient struc-
tures. Especially common are kites overlain by wheels,
which seem in some cases to be purposefully placed over
the ‘head’of the kite, perhaps as a deliberate act to
‘decommission’it (Fig. 17) (see Kennedy 2012c: 80), or
possibly just to take advantage of a cleared space. Some-
times there are pendants in close association, but more
commonly one ﬁnds pendants overlying wheels, implying
they are frequently the most recent of the three kinds of
structure (Fig. 18). It is rare to ﬁnd kites overlying any
other ancient structure except for another kite. One may
infer that of all the stone-built ancient structures in that
particular landscape, kites are among the very oldest—
wheels appear later and pendants last of all.
Such evidence is far less abundant in Harret Khaybar.
Despite the considerable number of stone-built structures
of several kinds, there are very few clear instances of a
kite intersecting with another structure apart from a
neighbouring kite. One reason may be the complete
Al-Wadi Kite 3 in the midst of monuments on a funerary avenue 5 km
west of Khaybar town (drawing: Rebecca Banks).
Samhah Kite 121 in a landscape of funerary monuments including a prominent avenue with several keyholes.
D. KENNEDY ET AL.
absence from this harra of the wheels which are found
in their thousands in Harret al-Shaam (see Kennedy
2012c). This lack of overlap is probably also due to the
different terrain favoured for kites and funerary features.
As has been seen, most kites are usually built near
wadis, often near the edges of the basalt lava ﬁeld (see
Fig. 26). In the immediate vicinity of the Khaybar and
Al-Ha’it oases, on the other hand, there are numerous
tracks radiating outwards, collectively ﬂanked by thou-
sands of pendants, keyhole structures and cairns. The
impression is of ‘funerary avenues,’with many of the
pendants and keyholes clearly oriented to these tracks.
Where they are built away from ‘avenues’, there is a
strong tendency for these structures to be built upon low
rises (as ascertained from the interpretation of geological
features through satellite imagery; such topography is
usually too small to be recognisable through freely avail-
able Digital Elevation Models). In both cases, it seems
that high visibility within the landscape is an important
factor in these features placement; the exact opposite of
what is likely to be true of kites. Nevertheless, several
rare juxtapositions of pendants and kites in this region
do provide a helpful indication of relative dating.
Two examples in Harret Khaybar may be illustrated; the
ﬁrst is As Silsilah Kite 50, which is overlain by As Silsilah
Pendant 6 (Fig. 19). The burial chamber of this very large
pendant might offer evidence allowing a terminus ante
quem for the kite. No pendant has yet been excavated and
dated in Harret Khaybar, but
C dates for three pendants
in the Yemen produced dates of 830, 630 and 60 BC (de
Maigret 2009: 329–35).
The second example is Al-Wadi Kite 3 (Fig. 20), which
is located within an extensive network of other funerary
stone structures aligned to an ‘avenue’c.5 km west of
Khaybar town. It seems unlikely that a kite would be con-
structed among so many existing obstructions.
There is a similar association in the Samhah area, with
Samhah Kite 121 situated in the midst of a dense land-
scape of funerary monuments—pendants and keyholes
(Fig. 21). It also seems likely here that the latter features
were built after the kite.
Also in the Samhah area (c.17 km north of Khaybar),
there is what seems to be a funerary avenue cutting
through the tail of Samhah Kite 62, again implying that
the kite is earlier (Fig. 22; note also Khaybar Kite 12,
Fig. 14). At the time these funerary landscapes were being
created, these kites were probably redundant.
Harret Khaybar is the major component of a group of three
coalesced lava ﬁelds, the other two being Harret Kura and
A tail of Samhah Kite 62 overlain (circled) by pendants of a funerary
avenue; (inset) detail of circled area.
The Mecca–Medina–Nafud volcanic line (from Roobol & Camp 1991:
KITES IN SAUDI ARABIA
Harret Ithnayn (Figs. 6 & 23). Together they cover an area
of 20,564 km
, with an estimated volume of c.1850 km
of lava (Roobol & Camp 1991: 2). They lie on the
c.600 km-long Mecca–Medina–Nafud volcanic line (Roo-
bol & Camp 1991; Camp, Roobol & Hooper 1991)
(Fig. 23). Harret Kura is the oldest at 10–5 million years
before the present (Ma); the other two are broadly coeval
at less than 5 Ma and less than 3 Ma respectively (Roobol
& Camp 1991: 2), but both contain lava ﬂows that are sig-
niﬁcantly younger, including potential dates into historic
Khaybar town is situated at around 750 m above sea
level (asl) and the land then rises gently eastwards to about
900–1000 m asl. The vent zone itself has peaks and other
landforms rising to about 2000 m asl. The oldest ﬂows
seem to be those that have travelled furthest and are found
Habir Flow revealed as black in the wider lava ﬁeld and showing the location of the kites discussed (Google Earth); (inset) Khaybar Kite 249 whose
tails may have originally terminated at the edges of the wadi.
A map of Habir Flow, Khaybar is on the left edge; the key area is Qb7 (pink) (extract from Roobol & Camp 1991).
D. KENNEDY ET AL.
around the periphery of the lava ﬁeld, where they have
been heavily eroded into a surface of rounded basalt boul-
ders. The youngest ﬂows are those appearing darkest in
satellite imagery (Fig. 24). Most notable is the Habir
Flow, narrow bands of molten lava, which have followed
wadi courses westwards to c.15–20 km south-east of
Khaybar town. Part of this ﬂow has been deﬁned geologi-
cally as ‘Qb6’and ‘Qb7’(Roobol & Camp 1991: 21–23)
(Fig. 25), dated as ‘post-Neolithic’and ‘historic’respec-
tively. Although they are visibly the most recent of the
lava ﬂows in this harra (not eroded, lacking dust-bowls,
etc.), the dating for Harret Khaybar is derived from the
traces of archaeological sites that it overlays. The geolo-
gists who mapped this area noted places where what
looked like parts of kites remained close by, and also some
kites apparently partially covered by the lava ﬂow (Camp,
Roobol & Hooper 1991: 368). Figure 24 shows ten kites,
parts of which seem to have been overlain by the Habir
Flow, that is, their tails originally terminated against the
edge of the wadis, the beds of which are now ﬁlled with
lava. One of these (Khaybar Kite 249; Fig. 24, inset) was
illustrated with a low oblique aerial photograph by the
geologists and cited as a clear example of an overlain
structure (Roobol & Camp 1991: ﬁg. 14).
Unfortunately, in this case the lava ﬂow is being dated
by archaeological structures, the age of which is highly
controversial. Until this lava ﬂow can be dated indepen-
dently, these examples from Harret Khaybar are potentially
interesting but in reality undated. They could be much
older than the Neolithic or indeed very much younger.
The two largest known groups of kites in Saudi Arabia are
those to the north in Harret al-Shaam and those in the
Distribution of kites and pendant features around the Al-Ha‘it oasis
(drawing: Matthew Dalton).
Distribution of gates; (left) in Harret Khaybar; (right) with other principal sites in the Khaybar study window (as of 01/12/2014) (drawing: Rebecca
KITES IN SAUDI ARABIA
central west of Harret Khaybar. There are several others
which may be brieﬂy noted. First are ﬁve probable kites
about 25 km north of Ha’il, the heads of which are invisi-
ble but the tails clear. None is on a harra. More signiﬁcant
are the eighteen kites in a single clustered group on the
lava ﬁeld along the north-eastern fringe of Harret Rahat.
Two features are noteworthy. Firstly, the wider landscape
of this harra is often available in high-resolution imagery
on virtual globes and numerous ‘works’of other kinds
have been recorded, but no further kites. Secondly, the
form of these kites is very unusual and generally quite dif-
ferent from those in Harret Khaybar. They appear as
‘heads’with ‘hides’along the edges of wadis but not one
has even a single tail. Such tailless kites are not unparal-
leled—there are a number of examples in Al-Ha’it (e.g.
Ash Shuwaymis Kites 2, 6?, 12, 16 and 21; Ar Rawd Kites
9, 14, 7 [and neighbours]; and al Ha’it Kite 1, etc). It is
notable that this small discrete group found in relative iso-
lation on the edge of a large lava ﬁeld are all of this type,
perhaps representing distinctive local hunting practices
and/or different fauna.
Despite the large number of kites identiﬁed in Harret
Khaybar, they are conﬁned overwhelmingly to two loca-
tions—the vicinities of the apparently long-inhabited oases
of Khaybar and Al-Ha’it—and they are never found far
beyond the limits of the lava ﬁeld. We may also note that
very few kites are found on Harret Kura, the oldest part of
the lava ﬁeld, and these are concentrated on its eastern
edge, closer to the Khaybar oasis. The inference that may
be drawn from the patterning of other stone-built struc-
tures (see Figs. 10 & 27) is that people certainly operated
in the interior of the harra but never built kites there.
Although this may be because kites, pendants, cairns, etc.
are each of different periods, it is more likely that their
absence reﬂects a lack of hunting opportunities in these
The overwhelming lack of visible stone structures on
the alluvial ﬂatlands adjacent to lava ﬁelds is unlikely to
reﬂect any subsequent covering by water-borne sediments;
the few features that have been built in these areas are usu-
ally still very clear. More prosaically, it seems likely that
these patterns can be accounted for by the need for struc-
tures to be built on or near appropriate sources of building
stone in the basalt boulder-strewn lava ﬁelds. Of course,
there is no a priori relationship between lava ﬁelds and
stone-built structures. No kites have yet been located in
Harret Ithnayn north of Al-Ha’it, for example, in spite of
some high-resolution coverage. Indeed, Harret Ithnayn—
insofar as it can be ‘seen’in satellite imagery—is almost
completely devoid of any visible man-made structures,
apart from a few very simple small corrals and a sparse
scatter of small pendants along its eastern edge.
Although work on the mapping of kites in Harret Khay-
bar is probably almost complete, it is unlikely that many
more will be found as the remaining parts of the harra are
covered by new high-resolution uploads and mapping
‘works’of other kinds is incomplete. Pendants of various
kinds are extremely common and cataloguing is
labour-intensive; cairns are more numerous still (and
simultaneously more difﬁcult to differentiate from natural
features). Nevertheless, some provisional observations can
be made about the relationship of kites to pendants both
for an incomplete map of the harra as a whole—as just
discussed—and for the area around Al-Ha’it oasis where
mapping is complete. While they usually occupy very dif-
ferent terrain (see above), densities of kites and pendant
features are broadly interrelated (Fig. 26); kites seldom
exist in isolation, and even in the most remote cases are
never further than 1 km from the nearest pendant.
Like pendants, there is also a tendency around Al-Ha’it
for the densest kite agglomerations to be found closer to
the oasis, with numbers generally becoming sparser as dis-
tance increases. One exception is the chain of kites situ-
ated on the north-west and south-east ﬂanks of a basalt
spur c.12–20 km north of the oasis; here it is especially
interesting to note that high kite densities mirror those of
an adjacent string of avenue-oriented pendants.
Taken together, the approximate correlation of these
features may indicate the more focused visitation of areas
closest to the oasis, and perhaps a decreasing willingness
to travel further from this point. The data for kites from
the original Khaybar window may also indicate this, with
kite concentration peaking at 4–8 km from the oasis, and
dropping drastically after 15 km. Could these patterns per-
haps indicate that ancient human occupation in this region
was centred on oases such as Khaybar and Al-Ha’it (as it
still is today)? The densely occupied mortuary landscapes
that radiate out from them certainly suggest a level of
inhabitation or at least important funerary signiﬁcance
One type of ‘work’for which mapping in Harret Khay-
bar is probably complete (or nearly so) is the one we have
called ‘gates’. This type of structure is not attested any-
where else, including Harret al-Shaam which has been
researched extensively and ﬂown over in almost every ﬂy-
ing season of the Aerial Archaeology in Jordan project
since 1998. They are found elsewhere in Saudi Arabia, but
the ﬁrst (and thus far largest) concentration is in Harret
D. KENNEDY ET AL.
Khaybar. In Figure 27 (left) it can be seen that gates occur
in the vicinity of the two towns but that there are also a
few in the harra interior, including—rather astonishingly
—ﬁve on the outer slopes of the bowl of one of the volca-
noes (Jabal al-Abyad). A total of 332 have been recorded.
In detail, when contrasting the pattern of kites with those
for gates and pendants (Fig. 27, right), it is evident that the for-
mer seldom mix with gates, while the latter occasionally over-
liethem.IntheareadeﬁnedbytheAl‘Ishash map sheet
(3926-32; Fig. 27, left), there are no less than thirty-three gates
—almost all of them on or just beyond the edge of the lava
ﬁeld—but only two kites (see Kennedy, in preparation).
As noted above, there is a correlation between kites and
watercourses (wadis), and it is no surprise that kites are
especially abundant in the vicinity of the oases of Khaybar
and Al-Ha’it. We can add, too, that a few kites (as in the
Al ‘Ishash map area) are found off the lava in sandy areas
reminiscent of the southern part of Harret al-Shaam (see
Fig. 3, cf. Fig. 4).
The kites seen in Harret Khaybar are not only notable for
their distinctive forms (especially the ‘barbed’type seen
above); types found elsewhere are absent. There are none
of the ‘sock head’type reported in Syria (Kennedy 2012b:
152–154) and south-west Jordan (Abu-Azizeh & Taraw-
neh, this volume), and perhaps that recorded by Ryckmans
(see Fig. 2). More striking is the absence of the immensely
complicated kites seen so frequently in Harret al-Shaam.
The latter are often kites that have been repeatedly modiﬁed
on essentially the same spot until the ﬁnal version appears
from above to have a complex ‘head’and multiple ‘tails’
(see Kennedy, Banks & Houghton 2014: passim). Like-
wise, the numerous examples in Harret al-Shaam of kites
that are plainly unﬁnished are found in just four instances
in the Harret Khaybar initial study window. Once again, it
is clear that the sequence of construction would be head,
tails, then hides. As noted above, it is rare for any kite in
Harret Khaybar to overlie/be overlain by any other structure
except another kite.
It is likely that within the next ﬁve years all of Harret
Khaybar will be covered by high-resolution imagery on
virtual globes. That will permit a ﬁnal tallying of kites,
although it is unlikely that many additional examples will
be revealed. Likewise, we can expect much more of Saudi
Arabia as a whole to receive such coverage and interpreta-
tion, allowing the presence and absence of kites across the
Peninsula to be determined. It is already apparent that kites
are not found in some harrat, or at least only in very small
numbers. Conversely, we may expect kites to be found
outside lava ﬁelds and perhaps to encounter kites of the
very unusual form found (so far) only in south-west
Yemen (Brunner 2009; cf. Kennedy 2011: 3193).
It remains the case that an effective and truly comprehen-
sive recording of kites and all of the other ‘Works of the
Old Men’visible in Harret Khaybar—and indeed, of Saudi
Arabia as a whole—should involve the systematic explora-
tion of the satellite imagery available through virtual
globes. This is feasible and the imagery is freely available.
Alongside that, however, should be the equally systematic
interpretation of whatever aerial photographs are held by
the Saudi authorities and by oil and mining companies.
That many of the latter will be old will be a particular
advantage in a country where rapid development has seen
dramatic clearance of the landscape for sprawling cities,
towns and villages, for infrastructure such as roads and for
agricultural development. Finally, as experience from eigh-
teen seasons of Aerial Archaeology in Jordan has under-
scored, satellite imagery and old survey photographs are
just part of a troika; ideally, a programme of low-level obli-
que aerial photography and the exploration of sites and fea-
tures invisible on satellite imagery would form the ﬁnal
strand. That in turn should involve—as is axiomatic for
programmes of aerial reconnaissance in Europe—ground
visits to explore the relationship between what is visible on
an image and what may be seen at ground level.
Following on from this last point would be a systematic
programme of ﬁeldwork by those with appropriate exper-
tise but guided by results and suggestions arising from this
preliminary exercise. Most obvious would be to investigate
those places where the juxtaposition of a kite and other
ancient structures may clarify and reﬁne chronology. It may
be possible to obtain absolute dates for sand deposits
beneath and against these structures by using the technique
of Optically Stimulated Luminescence dating. It would be
useful, too, as has been done in the Negev (Nadel et al.
2010) and now in south-east Jordan, to conduct exploratory
excavation on a kite (Abu-Azizeh & Tarawneh, this
volume) (cf. Brochier et al. 2014 and Crassard et al. 2014
for excavation on kites in Armenia and Kazakhstan).
Other places where speciﬁc kites should be investigated
on the ground are where they are associated with recent lava
ﬂows. The current programmes of research on the volcanic
activity and lava ﬁelds of Arabia may clarify dating of ﬂows
and sites affected by them (Moufti & N
emeth 2014; Moufti
et al. 2013; Murcia et al. 2014). It seems clear that the Ha-
bir Flow is relatively recent, Qb6 seems to overlie some of
these stone structures and Qb7 is younger still. There is
KITES IN SAUDI ARABIA
research in progress on such ﬂows through cosmogenic dat-
ing in Harret Rahat (just south-east of Medina), on ﬂows
previously dated by Camp and Roobol on the basis of asso-
ciated archaeological sites (Karoly N
emeth, personal com-
munication, 2 June 2014).
Finally, distribution maps of the type illustrated above
here should show kites within their environmental con-
texts of geology, soils, rainfall patterns and vegetation.
Appropriately high-resolution data probably already exist
but are not yet readily available.
Appendix: the other ‘works’in the Arabian lava ﬁelds
In addition to the large and familiar kites, other stone-built
ancient structures are:
wheel: a circular or roughly circular enclosure with a
central stone hub and radiating stone spokes. Some have an
external ring of small stone heaps; 20–50 m in diameter;
pendant: a substantial stone cairn from which stretches
out a tail, most commonly comprised of a line of small
stone heaps, or an elongated triangular enclosure. These
‘tails’of stone heaps can be as little as 10–15 m or as
large as 100 m or more;
keyhole: a type of pendant found only thus far in Harret
Khaybar, but in large numbers. From above they appear to
have the outline of an old-fashioned keyhole. The larger
can be as much as 50 m long, and often they are found
joined together in a series;
gate: a site type found only in Harret Khaybar and a
few other nearby areas. From above it resembles a fallen
barred gate. The longest is over 500 m but most are much
cairn: a heaped or constructed pile of stones. Function
can vary, but larger structures are associated with burials.
We are grateful to the Packard Humanities Institute, which
funded a pilot study of high-resolution imagery of
Harret Khaybar revealing the immense richness of this
Examples of the other ‘works’of the Arabian lava ﬁelds. From left to right, top to bottom: wheels, a pendant, keyholes and a gate (Google Earth).
D. KENNEDY ET AL.
Abu-Azizeh, W. & Tarawneh, M.B. this
volume 2015. Out of the harra: desert kites
in south-eastern Jordan. New results from
the South Eastern Badia Archaeological
Project. Arabian Archaeology and
Adams, R.McC, Parr, P.J., Ibrahim, M. &
Mughannum, A.S. 1977. Saudi Arabian
Archaeological Reconnaissance 1976.
Preliminary Report on the First Phase of
the Comprehensive Archaeological Survey
Program. Atlal 1: 21–40.
Betts, A. 2014. A response to Zeder, Bar-Oz,
Rufolo and Hole (2013). Quaternary
International 338: 125–127.
Betts, A.V.G. & Yagodin, V.N. 2000. A new
look at desert kites. Pages 31–43 in Stager,
L.E., Greene, J.A. & Coogan, M.D.
(eds.), The Archaeology of Jordan and
Beyond. Essays in Honor of James A.
Sauer. Winona Lake, IN: Eisenbrauns.
E., Barge, O., Karakhanyan, A. &
Kalantaryan, I. et al. 2014. Kites on the
margins. The Aragats kites in Armenia.
eorient 40/1: 25–53.
Brunner, U. 2009. Les pi
eges de chasse
antiques au Y
emen. Chroniques Y
Camp, V.E., Roobol, M.J. & Hooper, P.R.
1991. The Arabian continental Alkali
Basalt Province. 2. Evolution of Harrats
Khaybar, Ithnayn, and Kura, Kingdom of
Saudi Arabia. Geological Society of
America Bulletin 103/3: 363–391.
Crassard, R., Barge, O., Bichot, C.-E. &
E. et al. 2014. Addressing the
Desert Kites phenomenon and its global
range through a multi-proxy approach.
Journal of Archaeological Theory and
Method. Published online: 23 August
2014. Available at: http://
de Maigret, A. 2009. Arabia Felix. An
Exploration of the Archaeological History
of Yemen. (3rd edition). London: Stacey
Echallier, J.C. & Braemer, F. 1995. Nature et
fonctions des ‘Desert Kites’: donn
eses nouvelles. Pal
Kempe, S. & al-Malabeh, A. 2013. Desert
kites in Jordan and Saudi Arabia: Structure,
statistics and function, aGoogle Earth study.
Quaternary International 297: 126–146.
Kennedy, D.L. 2011. The ‘Works of the Old
Men’in Arabia. Remote sensing in interior
Arabia. Journal of Archaeological Science
Kennedy, D.L. 2012a. Pioneers Above Jordan.
Revealing a prehistoric landscape.
Antiquity 86: 474–491.
Kennedy, D.L. 2012b. Kites –New Discoveries
and a New Type. Arabian Archaeology and
Epigraphy 23/2: 145–155.
Kennedy, D.L. 2012c. Editorial: Wheels in the
Harret al-Shaam. Palestine Exploration
Quarterly 144/2: 77–81.
Kennedy, D.L. in preparation. A new site type
in Saudi Arabia –‘Gates’amongst the
‘Works of the Old Men’.
Kennedy, D.L. & Bishop, M.C. 2011. Google
Earth and the Archaeology of Saudi Arabia. A
case study from the Jeddah area. Journal of
Archaeological Science 38: 1284–1293.
Kennedy, D.L. & Al-Sa’eed, A. 2009. Desktop
Archaeology. Saudi Aramco World 60/4: 2–9.
Kennedy, D.L. with Banks, R. & Houghton, P.
2014. Kites in ‘Arabia’. [iBook version]
Emergent Form. Retrieved from http://
Moufti, R. & N
emeth, K. 2014. The White
Mountains of Harrat Khaybar, Kingdom of
Saudi Arabia. International Journal of Earth
Sciences (Geol Rundsch) 103: 1641–1643.
Moufti, M.R., N
emeth, K., Murcia, H. &
Lindsay, J.M. 2013. The 1256 AD Al
Madinah historic eruption geosite as the
youngest volcanic chain in the Kingdom
of Saudi Arabia. International Journal of
Earth Sciences (Geol Rundsch) 102:
Murcia, H., N
emeth, K., Moufti, M.R. &
Lindsay, J.M. et al. 2014. Late Holocene
lava ﬂow morphotypes of northern Harrat
Rahat, Kingdom of Saudi Arabia:
Implications for the description of
continental lava ﬁelds. Journal of Asian
Earth Sciences 84: 131–145.
Nadel, D., Bar-Oz, G., Avner, U., Boaretto, E.
& Malkinson, D. 2010. Walls, ramps and
pits: the construction of the Samar Desert
kites, southern Negev, Israel. Antiquity 84:
Parr, P.J., Harding, G.L. & Dayton, J.E. 1968–
69. Preliminary survey in N.W. Arabia,
1968. Bulletin of the Institute of
Archaeology (University of London) 8/9:
RCAHMS. Directorate of Overseas Surveys.
National Collection of Aerial Photography
(NCAP). Available at: http://ncap.org.uk/
(accessed 2 December 2014).
Roobol, M.J. & Camp, V.E. 1991. Explanatory
Notes to the Geologic Map of the
Cenozoic Lava Fields of Harrats Khaybar,
Ithnayn, and Kura, Kingdom of Saudi
Arabia (To accompany Map GM-131).
Riyadh: Ministry of Petroleum and
Ryckmans, J. 1976. An ancient stone structure
for the capture of Ibex in western Saudi
Arabia. Proceedings of the Seminar for
Arabian Studies 6: 161–165.
Zeder, M.A. & Bar-Oz, G. 2014. A response to
Betts (2014). Quaternary International
KITES IN SAUDI ARABIA