Isthmus Wall, II

In Galway, Scotland, there is a long and narrow valley called 'The Glen of the Bar' (55.007825° N, 4.379391° W).   It is about half a kilometer long and rapidly narrows from ~70 m to about 10 m at its head.

Figure 1. Glen of the Bar, facing NE to the valley head.

Figure 2.  Visitor's outlook.  Glen of the Bar, looking SW from valley head.

The Glen of the Bar was, in former days, reputed as a deer trap or 'elrick'.[1]  In this case deer would be driven by beaters up the narrowing valley until, at the last moment, they come over a rise and into an impound - up to this point invisible[2].  The corral or impound would have been just where the visitors' outlook is shown in figure 2.  An 'elrick' [3] is a deer trap consisting of a narrow defile (or even parallel walls) that narrows into an impound or corral in which game may be killed or captured (perhaps in nets).

  Communal hunting often makes use of narrow defiles or narrowing valleys to force game into such a tight spot that the animals can neither maneuver nor escape. Once animals are in that position they are easy prey for the hunter's arrow.

Besides the Glen of the Bar and other narrow defiles a classic case of game trap implementation is the use of lunate or parabolic sand dunes for trapping antelope. Once the animals are driven between the arms of the sand dune it is almost impossible for them to get out. The photograph shows why this is so. The sand dune shown here is some hundreds of meters long and, perhaps, 80 meters wide.

Figure 3.  A parabolic sand dune in the American South-West. [4]

Antelope driven into this dune will try to climb out. And yet if my readers look closely they will notice that, right at the edge of the dune, plant roots have held the sand nearly vertical for a distance of about two or three feet. It is very difficult for any animal to overcome this vertical lip. Animals exhaust themselves in the attempt and are killed. Many ungulates are superb jumpers and climbers but topography can be employed in such a way to negate those advantages.  In this sense they function just like the intermittent wall segments on the side of Mytika.

In the American west there often occur 'headcut' arroyos. These are defiles which are erosion cut by streams. In the process of their formation the upper end collapses into a vertical wall. The illustration shows what this looks like. Such arroyos are often the site of communal mass kills. [5] 

Figure 4.  A headcut arroyo.

 

La Roche de Solutré (46.299401° N, 4.719246° E) is a limestone outcropping in Burgundy, France.  I show it in the next picture.

 

Figure 4a. The Rock of Solutre (Roche de Solutré) in Burgundy.

For about 20,000 years it was the site of communal hunts of wild horses.  The actual site contains the remains of between 30,000 and 100,000 horses.  It is the premier site for prehistoric communal hunting in western Europe and proves that this form of hunting was understood as early as the Upper Paleolithic.[5a]  At one time it was believed that horses were herded up to the top of the rock and then forced to fall over killing them.  This thesis was debunked by some simple reflections on the behavior of horses.  Horses are not buffalo which travel in enormous herds.  Horses travel in small bands of several females led by a stallion.  The essence of the fall trap is a large number of animals which, by crowding, force those up front over the edge.  This is a situation which cannot be provoked among horses.

Instead it appears that when horses regularly passed the rock on migrations that men could spook them and change their direction  until they were trapped in a naturally occurring canyon.  Here they could be killed by waiting hunters.[5b]  The situation is hypothesized in the following picture.[5c]

What does this have to do with the valley between Mytika and Rachi?

We should start by facing facts.  No evidence has ever been produced to show that the 'Cyclopean' wall was intended to cross the Isthmus.  No evidence has ever been produced to indicate that the wall continued along the eastern side of the Rachi (about 970 m).  No evidence has ever been produced to indicate that the wall on the Mytika side was continuous and all we can conclude is that these segments were constructed ad hoc and, potentially, for differing purposes.  

A brook flows from the head of the valley and north-south through its entire length.  In antiquity this would have been a natural habitat for grazing ungulates [6] as well as many other types of game.  The flat valley bottom, gentle slope up to the head, the availability of water all allow us to suppose that this valley was a natural migratory route to and from the top of Mytika plateau where there would have been the abundant grassland for grazing.[7]  The area was "little inhabited" during the Bronze Age.  We know that the Mycenaeans were avid hunters of deer.  Many depictions of deer and deer-hunting figure in LH rings, cups, fresco, and pottery.

The entire valley between Mytika plateau and the Rachi is not inconsistent with a communal game-drive system.[8]  It bears a strong resemblance to the Glen of the Bar and many other such elricks.  That the segments Pe and Ge flare away from each other is a good indication that they were intended to be the beginnings of the 'antennae' or 'rays' of such a system.  The sequence Pe, Sp, Zo, Vl, and Pa steepen the W side of the Mytika plateau against which game would be driven. 

Figure 5.  The game-drive system in the Rachi-Mytika valley.  The 'Beaters' drive the game into the space between the 'rays' or 'antennae' at Mi-Ge and Pe.  'Stationary' refers to figures standing on the Rachci in outline against the sky.  Because of those figures the game will move further up into the valley and to the E where they encounter either the 'Shooters' or the 'Impound' where they may be captured or killed.  The 'Coast Road Ddefence' is a completely separate structure.

 If they were meant to contain deer then we should not expect to find any of these wall segments to be of a height of more than 3 m. which is more than high enough to defeat the jumping behavior of deer.[9]  When in use there would be several positions along the Mytika valley wall from which shooters would kill game.  This 'wall' need not be continuous along the whole line of the western Mytika slope.  On the steep Rachi side there are no walls nor would we expect to find them.  A line of 10 - 20 human figures standing against the skyline on the top of that ridge would be quite enough to drive game to the eastern side of the valley.  

Alternatively one would employ cairns with 'scares'[10] to accomplish the same purpose.  The fact that Mi only begins where the Rachi comes back to the level is highly indicative that the Mi-Ge sequence was the beginning of an 'antenna' or 'ray'.  Nothing that we know about the complex of wall segments in Broneer's thesis contradicts this model.

My game-drive system hypothesis might, of course, be false.  But it is far from absurd.  And it is far more likely than an hypothesis that attempts to relate archaeological finds to semi-mythological and highly suppositious 'events'.[11]

We know for a fact that the Mycenaeans were avid hunters and we know that they hunted the sort of game for which game drive systems are most efficient. Possible quarry for them would have included the aurochs (Bos primigenius) as well as the red deer (Cervus elaphus) which has been hunted in Greece during prehistoric as well as historic times. Other very likely game would be fallow deer (Dama dama) and roe deer (Capreolus capreolus). Literature and art also shows us that they were aficionados of boar hunting (Sus scrofa).  Nor must we forget the mainstay of the Mycenaean animal economy, ordinary cattle (Bos taurus).

Not very far from here the narrow valley at Klissoura[12] was used by Mesolithic people (long before greek-speakers came to Greece, of course) to trap and kill fallow deer which they seem to have done in large numbers. There are examples of this hunting style from about this same time in Epirus.[13]

Falsification

All good hypotheses should be falsifiable.  How would we attack this game-drive hypothesis?

a. The fastest way would be to support Broneer's hypothesis by discovering evidence that this was a cross-Isthmus wall intended for defensive purposes.  Additional fragments of the wall that indicated some clear direction might suffice.  It would also suffice to show defensive features: real towers, for example.  Such things would not be consistent with a game-drive system.  And, in order to be a protective wall, it need not have been completed.  There are other examples of partly completed walls on the Isthmus and those were certainly of a defensive nature.  [14]

   In fact I would not discount the probability of finding continuations of Mi-Ge.  But, if so, I think those segments would be found around the Sacred Glen - a prime candidate for a game-drive system in its own right.  Game-drive systems are often chained and that could be the case here.  See figure 6.

Figure 6.  Hypothetical game-drive system at the Sacred Glen.

b. As far as I am aware no one has proposed a convincing model of communal hunting anywhere else in Greece (or, at least, not in the BA).  'Absence of evidence' certainly would point in that direction.   A continuing dearth of such findings might significantly undercut my hypothesis.   But hunting in these societies leaves few material traces and is, perhaps, simply understudied or even, not studied at all.   One thing archaeologists could do is very simple - search for arrowheads at the foot of Sp, for example. [15] Another approach might be to re-examine the Mycenaean landscape for locations that would suggest hunting in the LBA.  The Klissoura valley wouldn't be the worst place to start.

c. The remains of Mycenaean feasting might show inordinately low amounts of wild game in the Mycenaean diet. This would definitely undermine my thesis.  But I suspect what researchers would find is that in the fully-developed Mycenaean system of LH III hunting was effectively restricted to the upper-most classes.  This sociological divide has been known in other places.

I regard the likelihood of any of criteria a through c as very unlikely to be realized but - maybe.

In the meantime I emphasize that my proposed model would explain, not only the 'missing wall' on the Rachi side,  but also the curious 'southern salient' in Broneer's proposed wall route.  In the game-drive model this 'southern salient' is no longer an anomaly. It is the whole point. 

  Dr. Broneer tried to base his hypothesis on the mythology of the 'Return of the Herakleidai'.[16]   Thinking about what the Mycenaean people ate would be a more productive approach.

Footnotes

[1] Fletcher [2011] pos. 8397.   An 'elrick' is a deer trap.  Fletcher, pos. 1583 says  "We begin to build up the picture of a gradation of deer enclosures from prehistory to the present, extending from the simple elrick or narrow defile into which deer might be driven for slaughter or capture, or through permanent or even temporary wings for capture in nets, to the enclosures into which deer might be enticed using browse such as ivy as previously discussed, to the fully enclosed ... deer park."

[2] Animal game drive corridors are almost always directed in such a way tthat the ultimate shooting positions, or the impound, etc. are invisible until the very last moment.  E.g. Kornfeld et al. [2016] 388: "Before the lane reaches the entrance to the corral, there is a deliberate bend, obscuring the corral entrance until the last possible moment ... ".  Mandelbaum [1940] 254, Fig. 5 illustrates a Cree buffalo impound which exhibits a similar sharp, last minute, bend, just before the impound itself.  Brochier et al. [2014] 28 emphasize that, of the upslope game drive traps he and his team studied in Armenia, 81% had such a slope break.

[3] Wiseman [2007] 244-5. " ... elerc seems to stem from Old Irish erelc, 'an ambush', which through metathesis later became in Gaelic either eileirig or iolairig, 'a deer-trap', i.e. a funnel shaped defile or V-shaped trap, either natural or artificial, into which deer were driven in order to be culled."

[4] Kornfeld  et al. [2016] 345-6.

[5] Ibid. 318. " Since headcuts in dry arroyos often formed barriers that prevented further movement upstream by bison, and the slopes of the adjacent walls of the arroyos were usually high enough and steep enough to contain the animals, these landforms were ideal for trapping the animals (Figure 4.6)."  There are many examples in Kornfeld et al. For example (p. 368) the Hawken site in Wyoming: "The Hawken site is a classic example of the arroyo bison trap; at least 80 animals were driven up the bottom of the arroyo by hunters until a perpendicular headcut was reached ... ".  And see the headcut trap in Agate Basin, Wyoming, pp. 318, 329, 338; the Carter-McGee site in the Powder River Basin in northern Wyoming, 330, 361; the Frasca and Nelson sites in NE Colorado, 362; and the Powder River again on p. 377, ""The banks of the arroyo at the time of the kill were believed to have been nearly vertical and 5–10 m high. The headcut that makes the trap was formed in more resistant strata in the Fort Union Formation ... "

 

[5a]  Olsen [1989] 295, "By the Upper Palaeolithic, there is considerable evidence that humans were very efficient communal hunters of large game."

[5b] Ibid., 316, 323.

[5c] Ibid., map: Fig. 2 from 297.

[6] Roe deer (Capreolus capreolus), Red deer(Cervus elaphus), Fallow deer (Dama dama), Aurochs (Bos primigenius), wild boar (Sus scrofa).

[7]  'natural migratory route'.  With the destruction of the deer in Greece we'll never know for sure.  Red deer survive in Greece only as a small herd on Parnitha.  The fallow deer survives in small herds on Rhodes.  There is no Capreolus in Greece to my knowledge.  The last aurochs died in Poland in 1627.  Wild boar survives in Greece but not, to my knowledge, in this area.

[8] For a brief introduction to such systems see this post.

[9] They don't need to be higher for game.   Deer are at a disadvantage jumping from below a wall. Red deer can clear 8 - 12 ft. (2.43 - 3.65 m) on the level but will only jump the greater heights when frightened or otherwise motivated (mating, food, alarm). This is true particularly if they have a running start.  Jumping much higher than that seems to put them at risk of injury.  Commercial estimates for required fence heights to constrain deer come in at somewhat less than that.  This is probably because the use of most commercial fences is merely to discourage deer and not keep them out under all circumstances (which can be significantly more expensive).  In a true game-drive system beaters would often be visible on the walls and the animals will instinctively shy back to the middle.

Fence height figures from howtorewild.co.uk:
Capreolus capreolus (roe deer): 150 cm. (4.92 ft.)
Cervus elaphus (red deer): 180 cm (5.9 ft.)
Dama dama (Fallow deer): 150 cm. (4.92 ft.)

[10]   'Scares'.  A common technique in the Arctic, for example.   Walls in the Arctic game-drive systems often are decorated with sticks to which fluttering flags are attached.  See Benedict [2005] 427.  The word in Latin is 'formido' which is defined as a rope strung with feathers used by hunters to scare game.  Hunters using nets with scares (formido) attached is described in Peck [1898] s.v. 'Rete' as '"This range of nets was flanked by cords, to which feathers dyed scarlet and of other bright colors were tied, so as to flare and flutter in the wind."  Fletcher [2011] pos. 2063 describes the Ring Hunt: " ...  once within a circle about the diameter of a league, about three miles, the animals were enclosed by a rope on which pieces of felt were hung. The hunters could then enter the circle and begin to kill them usually from horseback with bow and arrows."  Emphasis is mine.

[11] Broneer [1966] 357.  "These deductions are in agreement ... "  Whatever else we may say about Broneer's arguments (and, in fact, most arguments in this field) is that they are not deductions.

[12] For Klissoura: Stiner et al [2010]; Starkovich [2012].

[13] Epirus.  Sturdy et al. [1997] present a careful analysis of animal routes and habitats in Paleolithic Epirus.  They present (601) a model of animal capture and control in the area of the Asprochaliko gorge which greatly resembles this one.

[14]  Discussed in Wiseman [1963],  Wiseman [1978] 59-63.  The best and most recent survey may be in Gregory [1993] 4-6.  Gregory discusses not just the several walls but the strategic considerations which govern the defense of the Peloponnese from threats on the north.

[15] Arrowheads have been found at the ends of one or two 'Cursus' in Britain.  At Woldgate Cursus in Humberside, for example: 

Figure 7. Arrow heads found at one end of Woldgate Cursus.
Clearly this indicates the shooting position.

To my way of thinking this supports the overwhelmingly likely idea that the cursus was intended to drive game.  And see Fletcher [2011] pos. 1077 on the cursus: "I hesitate to suggest that they may ever have served as a means for killing deer ... but similar constructions were later to be used to direct deer."

[16] An example of the syndrome which I have named 'making Homer true'.  Archaeological finds can, perhaps, be used in responsible support of materials derived from epic literature or mythology (or vice versa).  A fine example of such responsible handling is in Robert Buck's History of Boeotia.when he comes to examine the founding myths of Thebes vs. what can be learned from archaeology and other sources.

Bibliography

Anderson [1985] : Anderson, J.K., Hunting in the Ancient World,  University of California Press, Berkeley, California. 1985. ISBN 0-520-05197-1.

Barringer [2001] : Barringer, Judith. The Hunt in Ancient Greece, The John Hopkins University Press, Baltimore and London. 2001. ISBN 0-8018-6656-1

Brochier et al. [2014] : Brochier, Jacques Élie with Olivier Barge, Christine Chataigner, Marie-Laure Chambrade, Arkadi Karakhanyan, Iren Kalantaryan, Frédéric Magnin. 'Kites on the margins. The Aragats kites in Armenia', Paléorient (40:1), pp. 25-53. 2014.

Buck, Robert [1979] : Robert Buck, A History of Boeotia.  University of Alberta Press, ISBN: 978-0888640512

Fletcher [2011] : Fletcher; John. Gardens of Earthly Delight: The History of Deer Parks. Windgather Press. Kindle Edition. 2011.

Gregory [1993] : Gregory, Timothy E., Isthmia V. The Hexamilion and the Fortress, American School of Classical Studies at Athens. Princeton, New Jersey. 1993. ISBN: 0-87661-935-9.

Kornfeld et al. [2016]³ : Kornfeld, Marcel with George C. Frison, and Mary Lou Larson. Prehistoric Hunter-Gatherers of the High Plains and Rockies, Routledge: Taylor & Francis Group. London and New York. 2016. ISBN 978-0-12-268561-3.

McOmish and Tuck [2002] : 'The Woldgate Cursus; Humberside, Survey Report', Archaeological Investigation Series. 2002.  Online here.

Olsen [1989] : Olsen, Sandra. 'Solutré: A theoretical approach to the reconstruction of Upper Palaeolithic hunting strategies', Journal of Human Evolution (18:4) pp. 295–327. 1989.

Starkovich [2012] : Starkovich, Britt M., 'Fallow Deer (Dama dama) Hunting During the Late Pleistocene at Klissoura Cave 1 (Peloponnese, Greece)', Mitteilungen der Gesellschaft für Urgeschichte (21), 11-36. 2012. Online here.

Stiner et al. [2010] : Stiner, Mary C. with Janusz K. Kozłowski, Steven L. Kuhn, Panagiotis Karkanas, and Margarita Koumouzelis, 'Klissoura Cave 1 and the Upper Paleolithic of Southern Greece in Cultural and Ecological Context', Eurasian Prehistory, (7:1) 309–321. 2010. Online here.

Sturdy et al. [1997] : Sturdy, Derek with Derrick Webley and Geoff Bailey. 'The palaeolithic geography of Epirus'. pgs. 587–614 in G. Bailey (ed.) Klithi: Palaeolithic Settlement and Quaternary Landscapes in Northwest Greece: Volume 2: Klithi in its local and regional setting. Cambridge: McDonald Institute for Archaeological Research. 1997. Online here.

Wiseman [2007] : Wiseman, Andrew E.M., Chasing the Deer: Hunting Iconography, Literature and Tradition of the Scottish Highlands, University of Ediburgh. 2007.  Online here.

Wiseman [1963] : Wiseman, James R., 'A Trans-Isthmian Fortification Wall', Hesperia; The Journal of the American School of Classical Studies at Athens [32:3], pp. 248-275, The American School of Classical Studies at Athens, 1963. Online here.

Wiseman [1978] : Wiseman, James. The Land of the Ancient Corinthians. Paul Åströms Förlag, Göteborg, Sweden. 1978. ISBN: 91-85058-78-5.

Isthmus Wall, I

The one step never taken in determining the purpose of the 'Cyclopean Wall' was to show that all the segments that supposedly constitute this wall actually are parts of a single structure. Broneer, Simpson, and others all merely assumed that this was true. [1]

Wiseman also accepted Broneer's interpretation of these segments as one structure but he does try to justify why they all should be considered the same structure. [2]

However, the sparse nature of these segments, their frustrating undatability, the cross-purposes at which they seem to operate, as well as the varying construction techniques and peculiar route of their hypothesized reconstruction cast strong doubt on this automatic assumption.[3] 

The discoverer of the first segment, Dr. Kardara, as well as the younger scholars, Morgan, Gregory, Loader, etc. weren't so easily convinced of the unity of this 'wall'. And while they have written clearly on this subject I see no convincing alternative explanations for the nature and location of these several segments.[4]

It is possible to see sections Ro, St, and Sk as having unified significance.[5] It seems that these three segments form a nearly straight-line stretching from the shore of the Saronic Gulf to a spur that leads up to the Mytika plateau. In other words this undoubted single wall (originally) proceeded from the beach, along a rise or bluff over a now-vanished creek, to the closest high ground. This group seems like a straightforward, lowlands coastal defensive wall.

Section Sk
Photo by Dr. Hugo Becker, 10/23.  All rights reserved.

From Ro it is 650 meters before we find the next segment in Broneer's scheme which is Pe. Of that route only about 230 m is shared with the Hexamilion wall.  So, in other words, after Ro, the 'wall' does not pick up again until we reach the other side of the top of the Mytika plateau.[6]

I suspect that Pe, Sp, Zo, Vl, and Pa form a complex of their own and divorced from the purposes of the previous. We might say the following things about this 'complex'.

A) Of all the segments in this group only Pe might realistically be part of a defensive wall based solely on the fact that it rests at or near the edge of the Mytika plateau. Broneer tried to emphasize this by suggesting that the forward extensions of Pe were 'small towers'. But the forward extensions of Pe are not 'towers'. Their small size and small extension (2/3 of a meter) are better interpreted as buttresses.[7]  Segment Pe does sit on moderately sloping ground and this lends support to the idea that these north-ward projections are simply buttresses.  So the attempt to make this stretch a defensive wall is weakened once the 'towers' are disposed of.

It is nearly a kilometer before any other segment (Pa) rises to the top of the terrain and this occurs, in my view, for very different reasons.

B) It is impossible that Sp ever formed part of a defensive wall. The top of Mytika here is about 82 m asl. Segment Sp is at 56 m asl. The difference is 26 m or 85 feet.[8]  In other words, behind Sp stretches a steep slope which is the height of an 8 story building. It simply cannot be that this segment was part of a defensive position.   It cannot be defended and the rule is that a wall that is not defended will fail.  And given the visible trend on both sides of Sp it seems that the course of the wall stayed well below the top of the ridge for a much greater distance.  Dr. Kardara, the discoverer of the first segments on Mytika, asserted her view that Sp was part of a buttress system for a road.[9]  As far as I am aware she seems never to have found a reason for changing her opinion about this.

C) Segment Zo is the odd man out of this group. It appears to have been built for purposes of its own, perhaps it was part of the much later Stadium attached to the Isthmian games. Photographs of Zo taken by an associate of mine dramatically demonstrate how very different its construction is from the other members of this group.[10]

Segment Zo (portion).
Photo by Dr. Hajo Becker. 10/2023. All rights reserved.

D) Vl appears to have been at about the same level as Sp. Not much is known about Vl (it has never been excavated) but, if its elevation was the same as Sp's (and there's no sign of it on the Mytika ridge above where I think it is), then again, it is impossible for it to have been part of a defensive system. I believe that this gives us some weak reason to think that Sp and Vl formed part of the same group. Simpson has suggested that it is obvious that this entire complex formed a defensive structure because it was constructed at the top of (Mytika) ridge. But between Pe and Pa this assertion is simply not true. No one ever built fortifications way BELOW the top of a ridge when a ridge was handy.[11]

E) Just after Vl the valley separating Rachi and Mytika rapidly narrows (from about 100 m to 50 m) and it rises to the level of the Mytika plateau where the Rachi joins it. At this juncture we find the wall segment Pa which seems to have been intended to join the Rachi side to a continuation (no longer to be seen) of Vl.   There is no wall continuation to the N on the Rachi side .

The head of the valley in vicinity of Segment Pa.  Photo faces N.
Rachi on left, Mytika plateau on right.
Photo by Dr. Hajo Becker. 10/2023. All rights reserved.

So, this group (if it is indeed a single group) starts at the edge of Mytika, sinks about 26 m in about 300 m to Sp and continues at this level (but above the valley bottom) until it rises again to end as Pa where the valley itself rises to meet at the juncture of the Mytika plateau and the Rachi. On the facing side of Rachi, for nearly a kilometer, nothing has been recovered until segment Mi. I include here a photograph taken partway down the valley and facing the valley mouth. This will give the clearest possible idea of the nature of the ground and how steep it is on both sides.

Rachi on L,  Mytika on R. 
Taken from  37.907818° N, 22.987935° E and facing N.

Photo by Dr. Hajo Becker. 10/2023. All rights reserved.

This map shows the position from which the previous photo was taken:

This bring us to our final group, Mi and Ge

F) The Mi and Ge sections form a group of their own. They are close together and are clearly parts of the same wall.  At present there is a 980 m gap between Pa and Mi.  In antiquity this gap may have been a bit smaller; sections of Mi seem to have been removed by local farmers. How much has been removed of what was originally built is anyone's guess.  When we plot Mi and Ge on a map we notice that they do not begin until the end of the Rachi has gone back to the valley floor.

There never was any continuation of the wall on the Rachi side to connect Pa and Mi.  In positing such a wall Broneer and Simpson lost sight of something fundamental.  It's not often enough remarked that the Mycenaeans were geniuses at getting hillsides to do their work for them. A good example of this is the draining of Lake Copais in Boeotia. In the thirteenth century BC the powers in charge at Orchomenos (some of this is suppositious but the dikes are the best evidence) wished to enlarge the channel of the Melas as a way to drain the Copais Lake.  (We do know that Lake Copais was drained under the Minyans.) The trick was to drain the Cephissus (a larger river than the Melas) into the channel of the Melas.  They actually did divert the Cephissos river in this way but found that the channel of the Melas wasn't large enough to handle the combined rivers.  So they built an enormous 17 km. dike from Orchomenos, just on the S edge of the Melas, and continuing to Topolia Bay on the east where the combined waters were drained into pre-existing sink holes or katavothrae (e.g., the Grand Katavothra at F857) which ring that bay. The point is that the Mycenaeans built nothing or very little on the north side of the Melas. They didn't need to. They simply let the plunging hills on that side be the other side of their channel. [12]

Segments Mi and Ge do not line up with the NE Rachi hillside but flare outward towards the W.  It was exactly this outward flare which caused Broneer to suppose that the wall might continue all the way across the Isthmus. And this flare to the W seems to find a mirror at Pe which flares eastward in the opposite direction. It is curious that these segments (about 300 m apart in a straight line) and an equal distance from Pa seem to form mirror images of each other.  What sort of walling activity would require that parallel walls should flare away from each other?

Whatever purpose led to the construction of Group 2 on the Mytika side - that purpose was fulfilled on the Rachi side by just utilizing the steep hills already existing and building little or nothing extra. And our hypothetical builders didn't build additional walls on that side until the Rachi itself terminated in the plain and became too low to carry out their purpose.

What purpose?

Footnotes

[1] Broneer [1966] 355.  "It is true that its line can be traced for only two kilometers from the sea, but there can be no doubt that it crossed-or was intended to cross-the Isthmus."  SImpson and Hagel [2006] 125: "Since the Wall has all the hallmarks of a single entity, it is not appropriate to treat it as if it was only a collection of incoherent sections." Simpson here would alllow an exception for segment Zo.  In footnote 63 here he seems uncharacteristically angry that Morgan [1999] doubts that these segments comprise one wall.

[2] Wiseman [1976] 59-60 gives a short description of the wall and his reasons for following Broneer.  And yet, in the course of this description, he makes several somewhat doubtful assumptions, the chief of which is that the projections of segment  Pe are 'A series of towers ... '.  From this assumption he is led ineluctably to the conclusion that the wall must be ' ... a fortification wall.'  Their small size makes it highly improbable that these projections were towers ... and in turn that makes the 'defensive wall' interpretation impossible.  

[3] The lengths of all the known segments, when added, amount to a little less than 10% of the hypothesized route of the wall.  As for the 'peculiar route' the chief problem in interpreting these segments remains the inexplicable plunge to the S after segment Pe and then the equally sudden reversal back to the N after Pa.  This adds a net 1500 m. to the length of the wall.  This 'southern salient' requires an explanation no matter how much Simpson tries to wish it away.  The datability question is addressed at length in Morgan [1999].

[4] Catherine Morgan [1999] provides a careful discussion of all segments of the wall.  She starts by saying "There is no evidence to indicate that it crossed, or was intended to cross, the Isthmus, or that it continued to Corinth and served to connect rural settlements; ... "  Loader [1995] 164-167 shows that some of the segments have widely varying dates.  Timothy Gregory [1993] 4 says " ... there is no evidence to connect these short sections into a great defensive work across the Isthmus; some of the sections might not be Mycenaean at all."

[5] For Sk, St, and Ro regarded as a group, see this.

[6] Between Ro and Pe there are two trial trenches, MW-3, MW-4, and a few scattered finds on the so-called 'Phytobanis' property.  Not enough information is preserved from these areas to support assertions that they form part of any cyclopean wall.  Morgan [1999] 442, nos. 3, 4, and 5.

[7] Even Broneer ([1966] 355) termed them 'miniature towers'.  And see Morgan [1999] 442-3: "Four projections, described as towers (although too small to be used effectively thus, and so perhaps buttresses), are spaced approximately evenly along the preserved, north, face, ... "

[8] A photograph of the steep slope behind Sp can be seen in Morgan [1999] 443, Fig. 8.

[9] Kardara [1971] (no page numbers but final page): "A retaining wall therefore for a road leading from the coast of the Saronic gulf to the main plateau of the Rachi - and hence to the mainland - must be reconsidered, inasmuch as this seems to be a more logical explanation than a defense wall across the Isthmos." And Gregory [1995] 5 "The wall on the slopes of Mytikas probably retained a road, ... "  Wiseman [1978] 60 attacks the idea by saying that the projections on section Pe are towers thus showing the wall's defensive purpose.  He also feels that it is illogical for a road to wind its way in and out of the ravine when it could travel by shorter routes.

[10]  In Blackman et al. [1997] 24 it is suggested that Section Zo may be associated with the later Hippodrome. "Based on the workmanship of the wall, portions of which contain ashlar masonry, it seems more likely that at least this portion of the wall is of Gr date."  Blackman means the Later Stadium.

    The area in and around Zo contains a complex series of ancient walls and it is not easy to determine if any of them are relevant cyclopean work.  Even Simpson felt that Zo might be an anomaly.  In Simpson and Hagel [2006] he says "All the sections of the Wall investigated by Broneer share the same basic characteristics ... with the sole exception of Section Zo."

 And Broneer suggested that "The terrain here slopes gently down along a small streambed, which runs almost straight north, and there may have been a road with a gate through the wall at this point. This could be the reason for the use of squared blocks."

[11] What little is known of Vl is summarized in Morgan [1999] 444: "Here the presence of a 30 meter long stretch of the north face is reported.  No further investigation or artifact collection was made."  See Broneer [1966] 352: "The outer face appears in several places, and one such stretch, on a property of Ioannis Vlassis, is 30 m. long."

[12] For use of cliffs to replace wall-building in game drive systems see Fradley et al. [2022] (online so no page numbers) when they come to discuss desert kites in western Saudi Arabia at Harrat Nawasif: " ..., and the use of slopes and cliffs as part of a natural enclosure. Cliffs were also integrated into the design of kites, possibly to reduce the amount of guiding walls that needed to be built ..."  For the channeling of the Cephissus river see this post.  

Bibliography

Blackman et al. [1997] : Blackman, David with Julian Baker and Nicholas Hardwick. 'Archaeology in Greece 1997-98', Archaeological Reports (44) 1-136. 1997. Online here.

Broneer [1966] : Broneer, Oscar.  "The Cyclopean Wall on the Isthmus of Corinth and Its Bearing on Late Bronze Age Chronology",  Hesperia: The Journal of the American School of Classical Studies at Athens (35:4), 346-362.  1966.  Online here.

Fradley et al. [2022] : Fradley, Michael with Francesca Simi and Maria Guagnin, 'Following the herds? A new distribution of hunting kites in Southwest Asia', The Holocene (32:11). pp. 1121-1131. 2022. Online here.

Gregory [1993] : Timothy, Gregory E., Isthmia V, The Hexamilion and the Fortress. American School of Classical Studies at Athens, Princeton, New Jersey, 1993.  ISBN: 0-87661-935-9.

Kardara [1971] : Kardara, Chrisoula. 'The Isthmian Wall; (A Retaining Wall for a Road)', Athens Annals of Archaeology (4:1), 85-89. 1971.  Online here.

Loader [1995] : Loader, Nancy. The definition of cyclopean: An investigation into the origins of the LH III fortifications on mainland Greece I, Durham theses, Durham University.. 1995. Online here.

Morgan [1999] : Morgan, Catherine.  Isthmia VIII; The Late Bronze Age Settlement and Early Iron Age Sanctuary. The American School of Classical Studies at Athens.  Princeton, New Jersey.  USA.  1999.  ISBN: 0-87661-938-3.

Simpson and Hagel [2006]: Simpson, R. Hope and D.K. Hagel, Mycenaean Fortifications, Highways, Dams and Canals.  Sävedalen 2006, Paul Åströms Förlag.  SIMA CXXXIII.   ISBN: 978-917081-212-5.

Wiseman [1978] : Wiseman, James. The Land of the Ancient Corinthians. Paul Åströms Förlag, Göteborg. ISBN: 91-85058-78-5.

The Banjo enclosure

 In my first post about hunting I discussed the so-called 'desert kites'.  These were game drive systems that consisted of long double lines of single stones (rarely amounting to so much as a primitive wall) with, every so often, a cairn or pile of stones.  At one end this double line was widely separated but the lines would gradually draw closer and at the narrowest end would lead into an impound corral or abut on a river, a lake or a fall where animals (in this case, gazelles) could be immobilized for killing.  I mentioned that this technique was world-wide, extending from the equator to the arctic, and was used by a wide variety of peoples for a wide variety of game (but most often ungulates).  

One thing that I didn't emphasize was that this game-drive method was only suitable for bare terrain: semi-arid plains, deserts, or tundra.  Game-drive systems using single-stone drive lines were rarely or ever used in more temperate climates, grassy plains, or, of course, in mixed wooded country.  This is simply because in such climate regimes single lines of stones would not be invisible. [1]  Community game drives and capture/killing were still performed but in modified form.

  I have been reviewing the literature on the 'banjo enclosure' which is characteristic of the British Isles during the late Iron Age (4th century BC to 1st century AD).  The banjo system consists of a roundish impound with game drive lines leading into it.  

The banjo enclosure at Sparrow's Copse.[2]
Sparrow's Copse is in Oxfordshire at about 51.540030° N, 1.413791° W
.

As the game drive 'antennae' or 'rays' are sometimes straight the entire assemblage resembles a banjo.  How are these constructed?  The banjo enclosure, including the antennae, consists of an inner bank with an outer ditch.  It has been suggested that as the banjo enclosure becomes larger (the average size is about 0.6 ha. but they may range as large as 2 ha.) this order is reversed having the ditch on the inside and the bank on the outside.[3]  For our purposes this doesn't much matter - the point is that the ditch and bank technique for creating the antennae and the corral is a modification of the 'stone lines' which I have been discussing up until now.  Scholars suggest that we should imagine the antennae banks as supplemented with hedges or blinds such that once animals are led into the drive lines their vision is sharply restricted.  

It's important to note this.   After about 2000 years the organic materials that completed a banjo enclosure are now disappeared.  What do I mean by 'organic materials'?  I mean the various blinds and 'scares' made of bushes, hedges, or hides have left no trace in the archaeological record.

One thing that is certain is that the bank and ditch antennae were, in some examples, extended to a considerable length with additional linear features. 

After reviewing this material I find it hard to come to any other conclusion than that this is a late Iron Age game-drive system.  The similarities are striking.  For example, we read this:

 "The entrance passages ... in areas where flint is common, they are paved with flint nodules." This is strongly reminiscent of the practice of the Amerindian impound at Muddy Creek in Wyoming. [4]  Why would these areas be paved with stone?  Because it makes the ground slippery for hooves and so the animal cannot spontaneously turn at the last minute - even if they were inclined to do so.

And there seems to have been an awareness that a height drop between the entrance antennae and the impound proper helped to prevent game from getting out of the impound.  The same technique is used in prehistoric America. [5]

Cult is involved as well.  Mandelbaum is concerned to emphasize the role that shamans play in communal game drives among the Cree indians.  A shamanic structure is often present at Cree game drives and this might go some way to explain the extra structures sometimes found in association with banjo enclosures. [6]  There are also indications that magico/religious practices were involved at the Ruby Impound site in Wyoming [7].

Another confirmation that this a classic game-drive system is that, in some of the examples I've seen, the antennae curve as the impound is approached so the animals cannot see where they are headed until the very last minute.

Between 150 and 200 of these banjo impounds are described.  They were discovered as a type only recently because the great majority are on level ground and are only visible as crop marks.  The thinking among scholars is that there must have been many more that will never be discovered due, among other things, to the destruction consequent to agricultural work. 

The researchers here provide interesting explanations of the function of these structures: McOmish says:

"..., many are now thought to be occupation sites, possibly of high status."[8]

I find it curious that the archaeologists on the spot are unable to formulate the game-drive hypothesis with respect to these structures.  It's one thing to suggest alternate hypotheses or to reject the game-drive hypothesis itself.  But the game-drive explanation for these structures is not even mentioned so that it can be dismissed.  One can only reach the conclusion that the researchers in the area of banjo enclosures have never heard of a hunting practice thousands of years old and used world-wide.

Footnotes

[1] Kornfeld et al. [2016] 400. "Even short grass can hide the drive lines during a normal grass year."

[2] Winton [2003] 20, Fig. 3.

[3] McOmish [2011/2018] 4.

[4] Flint nodules: McOmish, ibid.; at Muddy Creek impound in Wyoming.  Kornfeld et al. [2016] 391 say "An area of artificially placed boulders was found that created a pavement constructed at the edge of a flat area that forms part of the rim of the depression in which the corral lies."

[5] Height difference between entrance way and the impound floor. 

In McOmish, ibid, " ... at Church End Ring, Wiltshire, for example, the enclosure could only have been entered by way of steep ledge, such is the height difference between the entrance passage and the enclosure."   

In Kornfeld et al. [2016], "Postholes between the boulder pavement and the corral indicate a ramp was present; the animals were driven in the downstream direction onto the ramp, which dumped them into the corral and eliminated the need for an entrance gate."  

Mandelbaum sketches the design of such a ramp for the impound entrance constructed by Cree Indians.  

Mandelbaum [1940] 55, fig. 6: 

[6] Mandelbaum [1940] 52, 54.  " A pound had to be built under the supervision of a shaman who had been given the power to do so by a spirit helper." and "The shaman who directed the construction and operation of a pound (Fig. 5), chose the site in a thicket; ... "

[7] Kornfeld et al. [2016] 386-9. "The sophistication of the corral and related structures was matched or superseded by a religious structure alongside the final drive line. The identification of the remains as a religious structure is based on a number of features associated with the structure and the architecture of the structure itself."

[8] McOmish  [2011/2018] 6.

Bibliography

Kornfeld et al. [2016]³ : Kornfeld, Marcel with George C. Frison, and Mary Lou Larson. Prehistoric Hunter-Gatherers of the High Plains and Rockies, Routledge: Taylor & Francis Group. London and New York. 2016. ISBN 978-0-12-268561-3.

Mandelbaum [1940] : Mandelbaum, David G., The Plains Cree., Anthropological Papers Vol. 37, Pt. 2. American Museum of Natural History, New York. 1940. Online here.

McOmish [2011/2018] : McOmish, David. Banjo Enclosures; Introduction to Heritage Assets, Historic England. 2018. Online here.

Winton [2003] : Winton, Helen. 'Possible Iron Age 'Banjo' Enclosures on the Lambourn Downs', Oxoniensia (68), pp. 15-26.  Online here.