Why do the sediments tilt toward the slope in the soil just below the Uffington White Horse? Could it be because of

(for parts One and Two go here and here).

In 1990, as part of a wider program of excavation, a team of archaeologists from the Oxford Archaeology Unit (a charitable trust) got permission to dig two trenches (I’ll call them T1 and T2) across the Uffington Horse in an attempt to find both how its form had evolved since being first carved. Another trench was also dug in the valley bottom below the White Horse (called The Manger) to look for evidence of human occupation here. In 1994 two follow-up trenches (T3 and T4) were also dug around the White Horse in order to test some geophysical anomalies and also to try to date the monument.

All the trenches around the Horse showed that, where sampled, the Horse had changed little since it was first made, with only evidence of slight repositioning of the image further uphill, and for some levelling of the slope so that the horse was gradually becoming less visible from the valley. T3 allowed approximate dating of the horse’s first formation using Optically Stimulated Luminescence (OSL) dating. The trench in the Manger revealed only scattered artefacts but no occupation. The trenches around the Uffington Horse were intended to answer purely archaeological questions about the age and ancient form of the Horse. They were never intended to look for evidence of normal geomorphic processes; the people who drew the sections didn’t look for such evidence, and this is probably why complete sections are not published (the full records are held at the Oxfordshire County Museums Service and English Heritage in Swindon, both sadly now miles away from me). Despite these frustrations, it’s worth seeing what the published sections and the report can say to give weight to or to refute the White Horse landslip theory.

Map of the Uffington White Horse excavations of 1990 and 1994, showing the position of trenches T1 to T4 (in red) as well as the locations of the published cross-sections (in purple) (based on Miles et al. 2003).

Evidence from each trench

The Manger Trench: this revealed a very incomplete succession, with thick, periglacial Pleistocene debris overlain by thin Holocene layers until the Roman period, when rapid sedimentation seems to have occurred. One particular unit, ‘2009’, of probable end Pleistocene age, was described as being deposited under high energy conditions with blocks of chalk. What this means is difficult to know, but if it’s evidence of a landslip it’s probably too early for the White Horse.

T1: small and only partially published, this suggests that the Horse’s beak was first formed simply by packing chalk into a shallowly dug open pit or pre-existing hollow in the ground, as the silts below are not much disrupted.

T4: this is a trench with two spurs, of which a plan of the main trench and a section along one spur is published. It shows that if sediments ever accumulated on the slope above the horse they are now gone.

So much for T1, T4 and the Manger trench. However, trenches T2 and T3 are much more informative.

Sections T2 and T3a, redrawn from Miles et al. (2003), showing colluvium or soil layers overlying chalk bedrock. Yellow layers are coarse chalk rubble or ‘puddled’ chalk, buried former representations of the Uffington White Horse. The left-hand part of T3a is a guess, based on the descriptions given in the paper.

T2: this is 10m long, and a complete section is published for it. At the upper end of the trench archaeologists dug through a series of layers of ‘puddled chalk’, each representing a fresh image of the horse, often separated from each other by accumulated ‘colluvium’, the result of hillside erosion. The oldest of the ‘horse surfaces’ appeared to be created on a wide terrace, open down-slope but with a steeper face, coated in decayed chalk, behind it.

T3: this is a T-shaped trench, here divided into two (T3a and T3b). Only small parts of this trench have been published. This is a shame as the soil description for T3 is better and more detailed than that for T2. The published part of the T3a shows the same basic pattern as in the parallel T2. Interestingly, soil layers below the earliest ‘horse surface’ are described as often compacted, unlike all the soils above the surface, which are described as loose.

Section Analysis

In both sections T2 and T3a some layers show surfaces which dip up-slope (i.e. to the right). Normal deposition of sediment on slopes is with sediment layering dipping down the slope (as shown by almost all the black lines on the sections). Surfaces which dip up-slope can be for four reasons:

1) Rotational slippage has caused sediment layers to be rotated (clockwise on the sections).

2) Slippage of some form has created fault surfaces seen at various angles.

3) Natural erosion has created channels for the movement of sediment downslope.

4) Trenches have been excavated either by animals or people.

Sections T2 and T3a, as in the previous diagram, but with erosion surfaces and apparently upslope dipping layers highlighted as discussed in the main text.

In order to identify the cause of the up-slope dips I have highlighted certain features of sections T2 and T3a:

1) Bolder black lines show an obvious surface of erosion (revealed by the truncation of underlying layers) or excavation (e.g. the presence of ‘puddled’ chalk in a trough).

2) Blue lines mark those parts of the erosion/excavation surfaces which apparently dip upslope (to the right).

3) Red lines mark layers or surfaces which apparently dip upslope (to the right) but do not obviously appear to be erosion/excavation surfaces.

The two downslope sections reveal very similar patterns. In the upper layers of each section up-slope dipping surfaces are all or almost all erosion/excavation surfaces. It’s fairly safe to guess that much of this is the result of people cutting the ground to allow the insertion of chalk for each new horse surface, as was concluded by Miles et al. 2003. Some surfaces, such as on T3a, may also result from natural channel cutting by water.

The lower layers of each section, however include some surfaces dipping upslope which, on current evidence, don’t appear to be erosion/excavation or even fault traces. This leaves us with the possibility that their orientation is the result of landslip. Notably, all of these surfaces are older than – indeed just underlie – the earliest surface possible for the Uffington geoglyph.

The up-slope dips shown on these sections are apparent and the sections themselves do not give enough information to reveal the actual direction of dip. Certainly, as all are perpendicular to the slope direction then there must be some upslope component to the anomalous dips, even if it’s not straight up the slope. However, T3 was cut as two perpendicular ditches (T3a and T3b). By measuring the apparent dips for surfaces traceable in the sections of both ditches it’s possible using trigonometry* to calculate the true dip for some of the beds (NB this would have been easier if the full sections of the trenches had been published, but you take what you’re given).

All dips are found to be downslope except for surface B (5040/5023 in Miles et al. 2003), which lies just below the earliest horse surface. This layer dips Eastward at an angle of about 9°. The direction of dip, while not directly up the slope, is consistent with bedding rotation due to landslip with a main scarp in the form of the White Horse, although there may be an alternative explanation which I haven’t spotted.

Dip directions and dips (lengths of arrows) on various bedding surfaces from trench T3 (yellow dot). Surface B is anomalous in dipping upslope.

Thus the evidence that we have from the archaeological sections suggests, at least tentatively, that rotational landslip of the upper slope occurred at some time not that long before the formation of the White Horse and that there was little or no sediment deposited between these two events. This is about as near as I’m going to get to a smoking gun.


I’ve made my case for the Uffington White Horse as a landslip. It’s not watertight but it’s better than I thought it would be. What it now needs is for someone else, probably a geomorphologist or structural geologist, to look into this further.

What kind of landslip appears initially to be obvious. The evidence from the trenches suggests a rotational landslip. However, this is slightly difficult to square with the statement which I’ve had from Simon Palmer (one of the original archaeologists) that he believes the archaeologists dug down to chalk bedrock along all the four trenches around the White Horse.

Rotational slips often cut into the underlying bedrock to expose fresh bedrock, and this seems to be the most likely way to get a sharp chalk scar to appear on the Uffington hillside. However, the relatively smooth surface which marks the base of each trench shows no steps typical of faulting in the bedrock.

If the archaeologists really did dig down to bedrock in all sections then the only way to get a landslip would be a rotational slip within the colluvium only. This might expose a rusty yellow-brown chalk surface but it certainly wouldn’t be fresh and it might not have attracted significant attention when such a landslip occurred.

What kind of timescale such a slip appeared over is also difficult to estimate. It could have happened over months or years, or alternatively in one swift event. Either way, such a landslip must catch the attention of the local people or it would never be modified. The emergence of an animal shape on the downs would be an extraordinary sign from the gods, one who’s fame would spread.

I’m not going to overhype this, though. This is a little story. The subsequent history of the Uffington White Horse is going to be similar, whether created by people or nature. Its magic, either way, would be equally strong to later generations. However, if the Uffington White Horse is natural in origin then, strangely, it’s about as convincing a supernatural thing as one could ever encounter in the wilds of ancient Britain.


Barclay A. et al. 2003 Chapter 4: The Manger, Dragon Hill and the Barrows, p29-59. and …

Miles, D. et al. 2003 Chapter 5: The White Horse, In: “Uffington White Horse and Its Landscape: Investigations at White Horse Hill, Uffington, 1989–95, and Tower Hill, Ashbury, 1993–4” (Miles, D. et al. eds.), Oxford Archaeological Unit, p61-78.

(this book is now out of print and very expensive to get your hands on, but if you’re lucky a library might have it (thanks, Mark) and it contains the complete volume on a CD too).

Nash-Briggs, D. 2009 Reading the images on lron Age coins: 2. Horses of the day and night, Chris Rudd List 106, pp4.

Pollard, J. 2017 The Uffington White Horse geoglyph as sun-horse, Antiquity 91, 406–420.

Both of the articles above suggest that the Uffington Horse was a representation of the Sun Horse, which NW Europeans of the Iron Age believed dragged the sun around the sky.


Why a landslip is unlikely at the present day White Horse, but how it might have happened with enough rainfall in the past.

(Parts one and three are here and here).

In the first part of this multipost I explained why I originally thought that the Uffington White Horse might be a landslip. However, is a landslip honestly a realistic idea? As far as I can tell it’s never been suggested before for the White Horse, and perhaps there’s a very good reason for that. So in this post I want both to highlight the problems involved in postulating a landslip at Uffington, and to see whether those problems are insurmountable or not.

As a general rule, landslips are encouraged by three factors: increasing slope steepness, weakened or slippery subsoil, and increasing ground saturation by water. I’ll deal with each of these factors in turn:

Slope steepness

Slope steepness encourages landslip simply due to the effects of gravity.

Our horse is located, appropriately, on the edge of White Horse Hill, the highest point in the Berkshire Downs. It lies just above a break in slope at the head of a small but deep, NW facing valley, or coombe, called The Manger. This valley cuts the north facing chalk escarpment of the Berkshire Downs on the south side of the Thames Valley.

A second, similar coombe (here called “Britchcombe”) trends NE just to the east of The Manger. The heads of these two valleys have probably the steepest slopes in all of the Berkshire Downs, reaching almost 40° in places.

Digital Elevation Model of the Berkshire Downs and its north-facing escarpment. Higher areas are paler. I’ve enhanced the image to bring out steep slopes, which are shown by dark fringes. Here the gentle north escarpment is cut by occasional, steep-sided ‘coombes’ or valleys. The Manger and Britchcombe make up the very dark, and hence steep, pair of coombes with their heads almost touching at the top of the image. The north end of the ridge where they touch is White Horse Hill (USGS data).

Now breaks in slope on chalk escarpments are normally quite gentle and smooth as a result of a process known as ‘soil creep’. This causes the surface soil to inch down slope as the soil expands and contracts in alternating wet and dry conditions. What makes the breaks in slope around The Manger surprising is just how sharp they are despite the effects of soil creep, and this in itself is curious.

True scale section through the Manger and White Horse escarpment, showing the location of the white horse and of a trench (T2, in red) cut by the Oxford Archaeological Unit which is discussed in the next post (based on Ordnance Survey and OAU data).

The subsoil

Landslips tend to be more likely on loose soils or in weak, broken rocks. They are much more likely when these overlie clay layers which allow the layers above to slide as the clay can act as a slip plane.

The Uffington Horse lies in a landscape of thin ‘rendzina’ soil overlying chalk. Chalk is a weak rock with many fissures and cavities. Nevertheless, historically this has not produced great landslips, except on the coast where the undercutting of chalk cliffs leads to their failure, for example on the two sides of the English Channel/Manche, at Dover and Wissant.

Only in places where there is a sufficiently shallow, slippery clay layer underlying the chalk is it easy for landslips to occur inland. This is not the case below the White Horse. The Jurassic clay layers beneath it are too deep to have allowed them to encourage slippage in the chalk. This is an important negative point against the possibility of a landslip as a cause of the formation of the White Horse.

Water saturation

Water saturation can decrease the friction in soil or rock, allowing things to slide better by filling the spaces in the rock or soil with water and pushing the grains apart. In this, chalk is not the best thing to produce a landslip.

Chalk, a kind of limestone, is notorious for its high permeability, a result of the interconnected holes between individual grains, the numerous fissures where the chalk has been broken by gentle folding in the distant past, and the cavities that form in it when rain dissolves the carbonate that makes it up.

This means that it’s actually quite hard to build up enough water pressure to cause slope failure in chalk. Even though heavy rainfall has been implicated in increasing the chances of failure in chalk cliffs in France, it would take a lot more water to cause failure in the more gentle slopes of the Berkshire Downs.

Where does this leave us?

Under present conditions, the evidence is against a major landslip at Uffington. However, at this point I need to launch into a digression, because we need to explain how The Manger, the valley at the top of which Uffington White Horse sits, formed.

(Please feel free to skip this section if your short time on this world is too precious.)

A small digression: What formed the Manger?

The Manger and Britchcombe are both examples of “escarpment coombes”, a type of valley which cuts many chalk escarpments in southern England.

Valleys normally form through the action of liquid water or ice. The escarpment coombes of the Berkshire Downs have never contained glaciers, so liquid water must have been important in their formation. However, they currently contain no rivers, and even the spring line is at 110m above sea level, tens of metres below most of the valley floor. This lack of water is in part due to the high permeability of both the surface soil and of the underlying chalk.

In the past these valleys cannot have been so dry. In order to form, either the water table must have been much higher, or the ground was less permeable due, for example, to its being frozen (permafrost) during cold phases. For each of these scenarios there are two further possibilities: either the water flowed along the entire length of the valley as a river, cutting down as it went, or it emerged as a spring at the base of the escarpment, undercutting the rocks above (‘groundwater-sapping’) leading to localised collapse of the escarpment and cutting back of a valley.

This has led essentially to four different views of coombe formation:

1) Rivers cutting down when the water table was higher (perhaps in the Tertiary period).

2) Seasonal rivers (e.g. spring meltwater resulting from snow melt) cutting down when the ground was frozen (during glacial phases of the Quaternary).

3) Springs eroding and undercutting the escarpment back when the water table was higher (perhaps Tertiary again).

4) Springs eroding and undercutting the escarpment when thawing of permafrost released meltwater from the ground at the end of glacial phases (Quaternary again).

For our purposes it is safe to rule out both 1 and 2. The Manger and its adjacent coombe are cut back to near the top of the White Horse Hill, and the hilltop would never have produced enough surface water to have allowed the formation of two large river valleys.

Detailed elevation model showing White Horse Hill (pale) and the coombes of The Manger and Britchcombe cutting the hill to the North. The black outline indicates the contour around the hill top equal to the height of The Manger. As can be seen, the two coombes make unlikely river valleys, as they both appear to drain the summit of White Horse Hill.

This leaves us with spring sapping and headward erosion of the escarpment, either due to high water tables or the melting of permafrost. Both involve landslips in the formation of the valleys. If this is correct then landslips in both The Manger and Britchcombe must have happened in the past.

Return to rainfall

I’ve already said that a landslip in The Manger is unlikely under present conditions, but the section above suggests that they have happened here in the distant past. But could conditions have been different enough at the time of the formation of the White Horse, back in the late Bronze or early Iron Age?

Now the slope at this time wasn’t significantly steeper, the rocks were no different, and the land certainly wasn’t frozen due to glacial conditions. However, could the weather have been significantly wetter then – say, wet enough raise the water table to the top of the valley, or to temporarily saturate the chalk and cause it to fail? The answer is a possible, though not a definite, ‘yes’.

The Late Bronze Age and Early Iron Age climatic deterioration, more properly called the 2.8 kA BP event or the ‘Homeric Minimum’, occurred between around 850 and 550 BC when NW Europe experienced a ‘particularly wet and cold’ phase, unmatched in previous millennia, associated with very low Sun activity (the link is to do with protecting the Earth from cosmic rays, apparently, not the cooling of the Sun). River catchment data for the British lowlands also back this up, suggesting a major peak in flooding events between 850 and 750 BC. This may be a suitable event to have caused either the saturation of the ground or the raised water tables that are needed for a landslip.

The chances of such an event causing a landslip at Uffington are increased by human activity at the site. For much of the Holocene the slopes of chalk coombes were stabilised by the growth of scrub and trees. The removal of this tree cover by farmers from the Neolithic onward, coupled with the increasingly intensive use of the land for agriculture from the Bronze Age, is now thought to have caused considerable erosion of the chalk landscape. This can be seen in the accumulation of chalk-rich sediment (known as ‘colluvium’) on the slopes and in the bottom of chalk valleys from the Bronze Age until Roman times and beyond.

Whether this is enough to have caused a landslip in the Manger I don’t know, but the dates of the White Horse (1380 and 600 BC – 68% confidence) and the climatic deterioration overlap and are at least suggestive. Whatever, there is also a third piece of evidence to consider; the excavations of the White Horse and the Manger by the Oxford Archaeology Unit in 1990 and 1994. I’ll discuss these is in the third part of this post.


Ballantyne, C.K. & Harris, C. 1995 The Periglaciation of Great Britain, Cambridge, pp342 (p152-155).

Background to views on coombe formation.

Brown, A.G. 2008 The Bronze Age climate and environment of Britain, Bronze Age Review 1, 7-22.

Discussion of high rainfall in Britain during the early 1st millennium BC.

Cranfield University 2020 The Soils Guide.

This site details the kinds of soils found on the Berkshire Downs, including Upton 1 soils which are probably those underlying the White Horse.

Duperret, A. et al. 2004 Coastal chalk cliff instability in NW France: role of lithology, fracture pattern and rainfall, Geol. Soc. Lond. Engineering Geology Spec. Pub. 20, 33-55.

Laurenz, L., Ludecke, H.J. & Luning, S. 2019 Influence of solar activity on changes on European rainfall, J. Atmospheric & Solar-Terrestrial Phys. 185, 29-42.

Highlights a Grand Solar Minimum between approximately 810 BC and 610 BC.

Martin-Puertas, C. et al. 2012 Regional atmospheric circulation shifts induced by a grand solar minimum, Nature Geoscience 5, 397-401.

Small, R.J. 1964 The Escarpment Dry Valleys of the Wiltshire Chalk, Trans. Inst. Brit. Geog. 34, p33-52.

USGS Earth Explorer website.

Source for SRTM (Shuttle Radar Topography Mission) 1″ Digital Terrain Elevation Data of the Berkshire Downs.

Wilkinson, K. 2009 Regional Review of Geoarchaeology in the Southern Region: Colluvium, English Heritage Research Dept. Report Series 3/2009, pp48.

Discussion of chalk land use and soil erosion.


Uffington White Horse as a landslip (part 1) – 10 years on and TV

June 10, 2020

Almost exactly ten years ago I wrote a post, my first post, to stop me talking to my partner about an idea that was floating in my head. The idea was that the famed Uffington White Horse might originally have been a landslip scar, subsequently adapted into the animal shape that you now see. A […]

Read the full article →

Population collapse and chance

July 14, 2018

What do Alexander the Great’s march on Asia, the Arab Conquests and overly hard exams have in common? The answer is the increased chance of getting lucky when the numbers are low. So you’re sitting in an exam and the questions seem exceptionally hard compared to all those practice papers you were doing. How should […]

Read the full article →

Indo-Iranian language from the steppe it seems

April 1, 2018

Backing the wrong (or no) horse again! Just for anyone who hasn’t found it, there’s a preprint of a paper called, rather uglily, ‘The Genomic Formation of South and Central Asia’ on bioarxv. It’s all about the ancient DNA of Central Asia, Iran and northern Pakistan from the Bronze Age to the Iron Age. This […]

Read the full article →

The causes of language swapping and language shift

March 5, 2018

A review post outlining three main methods of language swapping: 1) Biggest language wins, 2) Richest or most violent language wins (this is helped if languages are similar), and 3) Everyone’s second language wins. This post is aimed at anyone who’s interested, although it comes from a particular discussion between me and others about the […]

Read the full article →

Words and rules, and the contrasting family trees of Indo-European

January 30, 2018

So, do the different linguistic family trees of Indo-European tell you different things? I think that grammar-based trees are better at telling you about the original splits between language sub-groups, whereas word-based trees can tell you about how isolated those sub-groups became. At the beginning of the millennium two papers were produced which both purported […]

Read the full article →

Proto-Indo-European homelands – ancient genetic clues at last?

November 12, 2017

(Neditors note – You know I now realise that so much of the argument below depends on my belief that elite dominance is not a sufficient mechanism for language change. I prefer to see major intrusion of a new language carrying people as necessary (i.e. that 30% or more of the population will now be […]

Read the full article →

No South American DNA in ancient Easter Island – and…?

November 4, 2017

Regardless of whether Rapanui ever made contact with South America, is the lack of ancient DNA evidence of South American contact in Easter Island that surprising? A very recent paper by Lars Fehren-Schmitz, Pontus Skoglund, and others, based on the ancient DNA (mitochondrial and nuclear) of five Easter Island / Rapanui individuals dated before 1700 […]

Read the full article →

European ancient DNA – the movie

October 1, 2017

We have learned loads from the autosomal DNA analyses of Europe’s ancient populations which have poured out of Harvard and other universities in the last two years. Europe was a restless place, changing people more than some of us would have guessed. Here’s the movie adaptation. I put together the video above for my own […]

Read the full article →